Hugging Face's logo

huggingCode11
/
NAVIG

Safetensors
Model card Files Community
huggingCode11 commited on
Commit
d7a9a0d
·
verified ·
1 Parent(s): 4e0eb60

Upload 36 files

Browse files
Files changed (36) hide show
  1. llava1_6-vicuna-7b-instruct/README.md +202 -0
  2. llava1_6-vicuna-7b-instruct/adapter_config.json +26 -0
  3. llava1_6-vicuna-7b-instruct/adapter_model.safetensors +3 -0
  4. llava1_6-vicuna-7b-instruct/additional_config.json +1 -0
  5. llava1_6-vicuna-7b-instruct/configuration.json +11 -0
  6. llava1_6-vicuna-7b-instruct/generation_config.json +7 -0
  7. llava1_6-vicuna-7b-instruct/optimizer.pt +3 -0
  8. llava1_6-vicuna-7b-instruct/rng_state.pth +3 -0
  9. llava1_6-vicuna-7b-instruct/scheduler.pt +3 -0
  10. llava1_6-vicuna-7b-instruct/sft_args.json +247 -0
  11. llava1_6-vicuna-7b-instruct/trainer_state.json +1084 -0
  12. llava1_6-vicuna-7b-instruct/training_args.bin +3 -0
  13. minicpm-v-v2_6-chat/README.md +202 -0
  14. minicpm-v-v2_6-chat/adapter_config.json +26 -0
  15. minicpm-v-v2_6-chat/adapter_model.safetensors +3 -0
  16. minicpm-v-v2_6-chat/additional_config.json +1 -0
  17. minicpm-v-v2_6-chat/configuration.json +13 -0
  18. minicpm-v-v2_6-chat/generation_config.json +7 -0
  19. minicpm-v-v2_6-chat/optimizer.pt +3 -0
  20. minicpm-v-v2_6-chat/rng_state.pth +3 -0
  21. minicpm-v-v2_6-chat/scheduler.pt +3 -0
  22. minicpm-v-v2_6-chat/sft_args.json +247 -0
  23. minicpm-v-v2_6-chat/trainer_state.json +1202 -0
  24. minicpm-v-v2_6-chat/training_args.bin +3 -0
  25. qwen2-vl-7b-instruct/README.md +202 -0
  26. qwen2-vl-7b-instruct/adapter_config.json +26 -0
  27. qwen2-vl-7b-instruct/adapter_model.safetensors +3 -0
  28. qwen2-vl-7b-instruct/additional_config.json +1 -0
  29. qwen2-vl-7b-instruct/configuration.json +14 -0
  30. qwen2-vl-7b-instruct/generation_config.json +11 -0
  31. qwen2-vl-7b-instruct/optimizer.pt +3 -0
  32. qwen2-vl-7b-instruct/rng_state.pth +3 -0
  33. qwen2-vl-7b-instruct/scheduler.pt +3 -0
  34. qwen2-vl-7b-instruct/sft_args.json +247 -0
  35. qwen2-vl-7b-instruct/trainer_state.json +1202 -0
  36. qwen2-vl-7b-instruct/training_args.bin +3 -0
llava1_6-vicuna-7b-instruct/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: /fs/clip-projects/geoguesser/vlms/llava/llava-v1.6-vicuna-7b-hf
3
+ library_name: peft
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+ ### Framework versions
201
+
202
+ - PEFT 0.12.0
llava1_6-vicuna-7b-instruct/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "/fs/clip-projects/geoguesser/vlms/llava/llava-v1.6-vicuna-7b-hf",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": [],
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "^(language_model|multi_modal_projector)(?!.*(lm_head|output|emb|wte|shared)).*",
23
+ "task_type": "CAUSAL_LM",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
llava1_6-vicuna-7b-instruct/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b66d04d5b5b9e30873fe7fbb4bd630ba8f28378484cdb0f6a6110e51a491d43
3
+ size 80446352
llava1_6-vicuna-7b-instruct/additional_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}
llava1_6-vicuna-7b-instruct/configuration.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "adapter_cfg": {
3
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/llava/llava-v1.6-vicuna-7b-hf",
4
+ "model_revision": "master",
5
+ "sft_type": "lora",
6
+ "tuner_backend": "peft",
7
+ "template_type": "llava-vicuna",
8
+ "dtype": "bf16",
9
+ "system": "A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions."
10
+ }
11
+ }
llava1_6-vicuna-7b-instruct/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "eos_token_id": 2,
4
+ "max_new_tokens": 2048,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.45.1"
7
+ }
llava1_6-vicuna-7b-instruct/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52792b98d009058a00eb4e56e3e140b84b88c4d07f4ba35a02deaa49a20f9351
3
+ size 161140126
llava1_6-vicuna-7b-instruct/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf85ab928113c942e29e418bc5b119c6dfca9cde55e4b30382d2ed66445b9953
3
+ size 14244
llava1_6-vicuna-7b-instruct/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b75bb0cab5c1ac64f1f42e99abb3e1c6d095394f6dd0c73c81aee621b476d6d9
3
+ size 1064
llava1_6-vicuna-7b-instruct/sft_args.json ADDED
@@ -0,0 +1,247 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "llava1_6-vicuna-7b-instruct",
3
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/llava/llava-v1.6-vicuna-7b-hf",
4
+ "model_revision": "master",
5
+ "full_determinism": false,
6
+ "sft_type": "lora",
7
+ "freeze_parameters": [],
8
+ "freeze_vit": false,
9
+ "freeze_parameters_ratio": 0.0,
10
+ "additional_trainable_parameters": [],
11
+ "tuner_backend": "peft",
12
+ "template_type": "llava-vicuna",
13
+ "output_dir": "/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625",
14
+ "add_output_dir_suffix": true,
15
+ "ddp_backend": null,
16
+ "ddp_find_unused_parameters": null,
17
+ "ddp_broadcast_buffers": null,
18
+ "ddp_timeout": 1800,
19
+ "seed": 42,
20
+ "resume_from_checkpoint": null,
21
+ "resume_only_model": false,
22
+ "ignore_data_skip": false,
23
+ "dtype": "bf16",
24
+ "packing": false,
25
+ "train_backend": "transformers",
26
+ "tp": 1,
27
+ "pp": 1,
28
+ "min_lr": null,
29
+ "sequence_parallel": false,
30
+ "model_kwargs": null,
31
+ "loss_name": null,
32
+ "dataset": [
33
+ "train_set_a.jsonl"
34
+ ],
35
+ "val_dataset": [
36
+ "test_set_a.jsonl"
37
+ ],
38
+ "dataset_seed": 42,
39
+ "dataset_test_ratio": 0.0,
40
+ "use_loss_scale": false,
41
+ "loss_scale_config_path": "/fs/clip-projects/geoguesser/zheyuan/DPO/swift/swift/llm/agent/default_loss_scale_config.json",
42
+ "system": "A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.",
43
+ "tools_prompt": "react_en",
44
+ "max_length": 2048,
45
+ "truncation_strategy": "delete",
46
+ "check_dataset_strategy": "none",
47
+ "streaming": false,
48
+ "streaming_val_size": 0,
49
+ "streaming_buffer_size": 16384,
50
+ "model_name": [
51
+ null,
52
+ null
53
+ ],
54
+ "model_author": [
55
+ null,
56
+ null
57
+ ],
58
+ "quant_method": null,
59
+ "quantization_bit": 0,
60
+ "hqq_axis": 0,
61
+ "hqq_dynamic_config_path": null,
62
+ "bnb_4bit_comp_dtype": "bf16",
63
+ "bnb_4bit_quant_type": "nf4",
64
+ "bnb_4bit_use_double_quant": true,
65
+ "bnb_4bit_quant_storage": null,
66
+ "rescale_image": -1,
67
+ "target_modules": "^(language_model|multi_modal_projector)(?!.*(lm_head|output|emb|wte|shared)).*",
68
+ "target_regex": null,
69
+ "modules_to_save": [],
70
+ "lora_rank": 8,
71
+ "lora_alpha": 32,
72
+ "lora_dropout": 0.05,
73
+ "lora_bias_trainable": "none",
74
+ "lora_dtype": null,
75
+ "lora_lr_ratio": null,
76
+ "use_rslora": false,
77
+ "use_dora": false,
78
+ "init_lora_weights": true,
79
+ "fourier_n_frequency": 2000,
80
+ "fourier_scaling": 300.0,
81
+ "rope_scaling": null,
82
+ "boft_block_size": 4,
83
+ "boft_block_num": 0,
84
+ "boft_n_butterfly_factor": 1,
85
+ "boft_dropout": 0.0,
86
+ "vera_rank": 256,
87
+ "vera_projection_prng_key": 0,
88
+ "vera_dropout": 0.0,
89
+ "vera_d_initial": 0.1,
90
+ "adapter_act": "gelu",
91
+ "adapter_length": 128,
92
+ "use_galore": false,
93
+ "galore_target_modules": null,
94
+ "galore_rank": 128,
95
+ "galore_update_proj_gap": 50,
96
+ "galore_scale": 1.0,
97
+ "galore_proj_type": "std",
98
+ "galore_optim_per_parameter": false,
99
+ "galore_with_embedding": false,
100
+ "galore_quantization": false,
101
+ "galore_proj_quant": false,
102
+ "galore_proj_bits": 4,
103
+ "galore_proj_group_size": 256,
104
+ "galore_cos_threshold": 0.4,
105
+ "galore_gamma_proj": 2,
106
+ "galore_queue_size": 5,
107
+ "adalora_target_r": 8,
108
+ "adalora_init_r": 12,
109
+ "adalora_tinit": 0,
110
+ "adalora_tfinal": 0,
111
+ "adalora_deltaT": 1,
112
+ "adalora_beta1": 0.85,
113
+ "adalora_beta2": 0.85,
114
+ "adalora_orth_reg_weight": 0.5,
115
+ "ia3_feedforward_modules": [],
116
+ "llamapro_num_new_blocks": 4,
117
+ "llamapro_num_groups": null,
118
+ "neftune_noise_alpha": null,
119
+ "neftune_backend": "transformers",
120
+ "lisa_activated_layers": 0,
121
+ "lisa_step_interval": 20,
122
+ "reft_layer_key": null,
123
+ "reft_layers": null,
124
+ "reft_rank": 4,
125
+ "reft_intervention_type": "LoreftIntervention",
126
+ "reft_args": null,
127
+ "use_liger": false,
128
+ "gradient_checkpointing": true,
129
+ "deepspeed": null,
130
+ "batch_size": 1,
131
+ "eval_batch_size": 1,
132
+ "auto_find_batch_size": false,
133
+ "num_train_epochs": 2,
134
+ "max_steps": -1,
135
+ "optim": "adamw_torch",
136
+ "adam_beta1": 0.9,
137
+ "adam_beta2": 0.95,
138
+ "adam_epsilon": 1e-08,
139
+ "learning_rate": 0.0001,
140
+ "weight_decay": 0.1,
141
+ "gradient_accumulation_steps": 4,
142
+ "max_grad_norm": 1,
143
+ "predict_with_generate": false,
144
+ "lr_scheduler_type": "cosine",
145
+ "lr_scheduler_kwargs": {},
146
+ "warmup_ratio": 0.05,
147
+ "warmup_steps": 0,
148
+ "eval_steps": 50,
149
+ "save_steps": 50,
150
+ "save_only_model": false,
151
+ "save_total_limit": 2,
152
+ "logging_steps": 5,
153
+ "acc_steps": 1,
154
+ "dataloader_num_workers": 1,
155
+ "dataloader_pin_memory": true,
156
+ "dataloader_drop_last": false,
157
+ "push_to_hub": false,
158
+ "hub_model_id": null,
159
+ "hub_token": null,
160
+ "hub_private_repo": false,
161
+ "hub_strategy": "every_save",
162
+ "test_oom_error": false,
163
+ "disable_tqdm": false,
164
+ "lazy_tokenize": true,
165
+ "preprocess_num_proc": 1,
166
+ "use_flash_attn": null,
167
+ "ignore_args_error": false,
168
+ "check_model_is_latest": true,
169
+ "logging_dir": "/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625/runs",
170
+ "report_to": [
171
+ "tensorboard"
172
+ ],
173
+ "acc_strategy": "token",
174
+ "save_on_each_node": false,
175
+ "evaluation_strategy": "steps",
176
+ "save_strategy": "steps",
177
+ "save_safetensors": true,
178
+ "gpu_memory_fraction": null,
179
+ "include_num_input_tokens_seen": false,
180
+ "local_repo_path": null,
181
+ "custom_register_path": null,
182
+ "custom_dataset_info": null,
183
+ "device_map_config": null,
184
+ "device_max_memory": [],
185
+ "max_new_tokens": 2048,
186
+ "do_sample": null,
187
+ "temperature": null,
188
+ "top_k": null,
189
+ "top_p": null,
190
+ "repetition_penalty": null,
191
+ "num_beams": 1,
192
+ "fsdp": "",
193
+ "fsdp_config": null,
194
+ "sequence_parallel_size": 1,
195
+ "model_layer_cls_name": null,
196
+ "metric_warmup_step": 0,
197
+ "fsdp_num": 1,
198
+ "per_device_train_batch_size": null,
199
+ "per_device_eval_batch_size": null,
200
+ "eval_strategy": null,
201
+ "self_cognition_sample": 0,
202
+ "train_dataset_mix_ratio": 0.0,
203
+ "train_dataset_mix_ds": [
204
+ "ms-bench"
205
+ ],
206
+ "train_dataset_sample": -1,
207
+ "val_dataset_sample": null,
208
+ "safe_serialization": null,
209
+ "only_save_model": null,
210
+ "neftune_alpha": null,
211
+ "deepspeed_config_path": null,
212
+ "model_cache_dir": null,
213
+ "lora_dropout_p": null,
214
+ "lora_target_modules": [],
215
+ "lora_target_regex": null,
216
+ "lora_modules_to_save": [],
217
+ "boft_target_modules": [],
218
+ "boft_modules_to_save": [],
219
+ "vera_target_modules": [],
220
+ "vera_modules_to_save": [],
221
+ "ia3_target_modules": [],
222
+ "ia3_modules_to_save": [],
223
+ "custom_train_dataset_path": [],
224
+ "custom_val_dataset_path": [],
225
+ "device_map_config_path": null,
226
+ "push_hub_strategy": null,
227
+ "use_self_cognition": false,
228
+ "is_multimodal": true,
229
+ "is_vision": true,
230
+ "lora_use_embedding": false,
231
+ "lora_use_all": false,
232
+ "lora_m2s_use_embedding": false,
233
+ "lora_m2s_use_ln": false,
234
+ "torch_dtype": "torch.bfloat16",
235
+ "fp16": false,
236
+ "bf16": true,
237
+ "rank": -1,
238
+ "local_rank": -1,
239
+ "world_size": 1,
240
+ "local_world_size": 1,
241
+ "bnb_4bit_compute_dtype": "torch.bfloat16",
242
+ "load_in_4bit": false,
243
+ "load_in_8bit": false,
244
+ "train_sampler_random": true,
245
+ "train_type": "sft",
246
+ "training_args": "Seq2SeqTrainingArguments(output_dir='/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=4, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0001, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1, num_train_epochs=2, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs={}, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=5, logging_nan_inf_filter=True, save_strategy=<IntervalStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=2, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=50, dataloader_num_workers=1, dataloader_prefetch_factor=None, past_index=-1, run_name='/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed=None, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=False, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy=None, push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=False, include_num_input_tokens_seen=False, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=GenerationConfig {\n \"bos_token_id\": 1,\n \"eos_token_id\": 2,\n \"max_new_tokens\": 2048,\n \"pad_token_id\": 0\n}\n, acc_strategy='token', loss_name=None, additional_saved_files=[], train_sampler_random=True, metric_warmup_step=0, train_dataset_sample=-1)"
247
+ }
llava1_6-vicuna-7b-instruct/trainer_state.json ADDED
@@ -0,0 +1,1084 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.41787565,
3
+ "best_model_checkpoint": "/fs/clip-projects/geoguesser/vlms/llava/output/llava1_6-vicuna-7b-instruct/v10-20241108-045625/checkpoint-534",
4
+ "epoch": 1.9962616822429906,
5
+ "eval_steps": 50,
6
+ "global_step": 534,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.003738317757009346,
13
+ "grad_norm": 0.7382091283798218,
14
+ "learning_rate": 3.7037037037037037e-06,
15
+ "loss": 1.75865197,
16
+ "memory(GiB)": 21.51,
17
+ "step": 1,
18
+ "train_speed(iter/s)": 0.03225
19
+ },
20
+ {
21
+ "epoch": 0.018691588785046728,
22
+ "grad_norm": 0.7008568048477173,
23
+ "learning_rate": 1.8518518518518518e-05,
24
+ "loss": 1.8970871,
25
+ "memory(GiB)": 21.51,
26
+ "step": 5,
27
+ "train_speed(iter/s)": 0.107257
28
+ },
29
+ {
30
+ "epoch": 0.037383177570093455,
31
+ "grad_norm": 0.6195642948150635,
32
+ "learning_rate": 3.7037037037037037e-05,
33
+ "loss": 1.85505066,
34
+ "memory(GiB)": 26.65,
35
+ "step": 10,
36
+ "train_speed(iter/s)": 0.150961
37
+ },
38
+ {
39
+ "epoch": 0.056074766355140186,
40
+ "grad_norm": 0.8053833842277527,
41
+ "learning_rate": 5.555555555555556e-05,
42
+ "loss": 1.85264435,
43
+ "memory(GiB)": 26.65,
44
+ "step": 15,
45
+ "train_speed(iter/s)": 0.174044
46
+ },
47
+ {
48
+ "epoch": 0.07476635514018691,
49
+ "grad_norm": 0.9945815205574036,
50
+ "learning_rate": 7.407407407407407e-05,
51
+ "loss": 1.75669136,
52
+ "memory(GiB)": 26.65,
53
+ "step": 20,
54
+ "train_speed(iter/s)": 0.188377
55
+ },
56
+ {
57
+ "epoch": 0.09345794392523364,
58
+ "grad_norm": 1.0137534141540527,
59
+ "learning_rate": 9.25925925925926e-05,
60
+ "loss": 1.5954113,
61
+ "memory(GiB)": 26.65,
62
+ "step": 25,
63
+ "train_speed(iter/s)": 0.198196
64
+ },
65
+ {
66
+ "epoch": 0.11214953271028037,
67
+ "grad_norm": 1.0105948448181152,
68
+ "learning_rate": 9.999136119166803e-05,
69
+ "loss": 1.67307549,
70
+ "memory(GiB)": 26.65,
71
+ "step": 30,
72
+ "train_speed(iter/s)": 0.205137
73
+ },
74
+ {
75
+ "epoch": 0.1308411214953271,
76
+ "grad_norm": 1.0798794031143188,
77
+ "learning_rate": 9.99385792841537e-05,
78
+ "loss": 1.68129864,
79
+ "memory(GiB)": 26.65,
80
+ "step": 35,
81
+ "train_speed(iter/s)": 0.210084
82
+ },
83
+ {
84
+ "epoch": 0.14953271028037382,
85
+ "grad_norm": 1.0679413080215454,
86
+ "learning_rate": 9.983786540671051e-05,
87
+ "loss": 1.61183624,
88
+ "memory(GiB)": 26.65,
89
+ "step": 40,
90
+ "train_speed(iter/s)": 0.214094
91
+ },
92
+ {
93
+ "epoch": 0.16822429906542055,
94
+ "grad_norm": 0.9876216053962708,
95
+ "learning_rate": 9.968931622637652e-05,
96
+ "loss": 1.5409358,
97
+ "memory(GiB)": 26.65,
98
+ "step": 45,
99
+ "train_speed(iter/s)": 0.217302
100
+ },
101
+ {
102
+ "epoch": 0.18691588785046728,
103
+ "grad_norm": 0.979777991771698,
104
+ "learning_rate": 9.949307432339625e-05,
105
+ "loss": 1.60590649,
106
+ "memory(GiB)": 26.65,
107
+ "step": 50,
108
+ "train_speed(iter/s)": 0.219869
109
+ },
110
+ {
111
+ "epoch": 0.18691588785046728,
112
+ "eval_loss": 1.543888807296753,
113
+ "eval_runtime": 18.3158,
114
+ "eval_samples_per_second": 2.73,
115
+ "eval_steps_per_second": 2.73,
116
+ "step": 50
117
+ },
118
+ {
119
+ "epoch": 0.205607476635514,
120
+ "grad_norm": 1.0606234073638916,
121
+ "learning_rate": 9.924932805436949e-05,
122
+ "loss": 1.54525614,
123
+ "memory(GiB)": 26.65,
124
+ "step": 55,
125
+ "train_speed(iter/s)": 0.206189
126
+ },
127
+ {
128
+ "epoch": 0.22429906542056074,
129
+ "grad_norm": 1.0801302194595337,
130
+ "learning_rate": 9.895831137146318e-05,
131
+ "loss": 1.54319582,
132
+ "memory(GiB)": 26.65,
133
+ "step": 60,
134
+ "train_speed(iter/s)": 0.209015
135
+ },
136
+ {
137
+ "epoch": 0.24299065420560748,
138
+ "grad_norm": 1.0459623336791992,
139
+ "learning_rate": 9.862030359785981e-05,
140
+ "loss": 1.55986643,
141
+ "memory(GiB)": 26.65,
142
+ "step": 65,
143
+ "train_speed(iter/s)": 0.211483
144
+ },
145
+ {
146
+ "epoch": 0.2616822429906542,
147
+ "grad_norm": 1.0878509283065796,
148
+ "learning_rate": 9.82356291596578e-05,
149
+ "loss": 1.54775982,
150
+ "memory(GiB)": 26.65,
151
+ "step": 70,
152
+ "train_speed(iter/s)": 0.213594
153
+ },
154
+ {
155
+ "epoch": 0.2803738317757009,
156
+ "grad_norm": 1.0929535627365112,
157
+ "learning_rate": 9.780465727448149e-05,
158
+ "loss": 1.60084972,
159
+ "memory(GiB)": 26.65,
160
+ "step": 75,
161
+ "train_speed(iter/s)": 0.215384
162
+ },
163
+ {
164
+ "epoch": 0.29906542056074764,
165
+ "grad_norm": 1.0857256650924683,
166
+ "learning_rate": 9.732780159709912e-05,
167
+ "loss": 1.53291664,
168
+ "memory(GiB)": 26.65,
169
+ "step": 80,
170
+ "train_speed(iter/s)": 0.217022
171
+ },
172
+ {
173
+ "epoch": 0.3177570093457944,
174
+ "grad_norm": 1.0876630544662476,
175
+ "learning_rate": 9.680551982238942e-05,
176
+ "loss": 1.49946527,
177
+ "memory(GiB)": 26.65,
178
+ "step": 85,
179
+ "train_speed(iter/s)": 0.218544
180
+ },
181
+ {
182
+ "epoch": 0.3364485981308411,
183
+ "grad_norm": 1.0945876836776733,
184
+ "learning_rate": 9.623831324603754e-05,
185
+ "loss": 1.57413607,
186
+ "memory(GiB)": 26.65,
187
+ "step": 90,
188
+ "train_speed(iter/s)": 0.219824
189
+ },
190
+ {
191
+ "epoch": 0.35514018691588783,
192
+ "grad_norm": 0.9552567601203918,
193
+ "learning_rate": 9.562672628338233e-05,
194
+ "loss": 1.47238646,
195
+ "memory(GiB)": 26.65,
196
+ "step": 95,
197
+ "train_speed(iter/s)": 0.221014
198
+ },
199
+ {
200
+ "epoch": 0.37383177570093457,
201
+ "grad_norm": 1.0762443542480469,
202
+ "learning_rate": 9.497134594687634e-05,
203
+ "loss": 1.60602245,
204
+ "memory(GiB)": 26.65,
205
+ "step": 100,
206
+ "train_speed(iter/s)": 0.222132
207
+ },
208
+ {
209
+ "epoch": 0.37383177570093457,
210
+ "eval_loss": 1.4929084777832031,
211
+ "eval_runtime": 14.0965,
212
+ "eval_samples_per_second": 3.547,
213
+ "eval_steps_per_second": 3.547,
214
+ "step": 100
215
+ },
216
+ {
217
+ "epoch": 0.3925233644859813,
218
+ "grad_norm": 0.985505998134613,
219
+ "learning_rate": 9.42728012826605e-05,
220
+ "loss": 1.53017511,
221
+ "memory(GiB)": 26.65,
222
+ "step": 105,
223
+ "train_speed(iter/s)": 0.216155
224
+ },
225
+ {
226
+ "epoch": 0.411214953271028,
227
+ "grad_norm": 1.0371544361114502,
228
+ "learning_rate": 9.353176276679396e-05,
229
+ "loss": 1.55461969,
230
+ "memory(GiB)": 26.65,
231
+ "step": 110,
232
+ "train_speed(iter/s)": 0.217382
233
+ },
234
+ {
235
+ "epoch": 0.42990654205607476,
236
+ "grad_norm": 1.1553157567977905,
237
+ "learning_rate": 9.274894166171888e-05,
238
+ "loss": 1.53458586,
239
+ "memory(GiB)": 26.65,
240
+ "step": 115,
241
+ "train_speed(iter/s)": 0.218618
242
+ },
243
+ {
244
+ "epoch": 0.4485981308411215,
245
+ "grad_norm": 1.062723994255066,
246
+ "learning_rate": 9.192508933357753e-05,
247
+ "loss": 1.56342993,
248
+ "memory(GiB)": 26.65,
249
+ "step": 120,
250
+ "train_speed(iter/s)": 0.21963
251
+ },
252
+ {
253
+ "epoch": 0.4672897196261682,
254
+ "grad_norm": 0.9741066098213196,
255
+ "learning_rate": 9.106099653103728e-05,
256
+ "loss": 1.46541033,
257
+ "memory(GiB)": 26.65,
258
+ "step": 125,
259
+ "train_speed(iter/s)": 0.220578
260
+ },
261
+ {
262
+ "epoch": 0.48598130841121495,
263
+ "grad_norm": 1.1155296564102173,
264
+ "learning_rate": 9.015749262631536e-05,
265
+ "loss": 1.45173082,
266
+ "memory(GiB)": 26.65,
267
+ "step": 130,
268
+ "train_speed(iter/s)": 0.221485
269
+ },
270
+ {
271
+ "epoch": 0.5046728971962616,
272
+ "grad_norm": 1.3632838726043701,
273
+ "learning_rate": 8.921544481913218e-05,
274
+ "loss": 1.51770496,
275
+ "memory(GiB)": 26.65,
276
+ "step": 135,
277
+ "train_speed(iter/s)": 0.222401
278
+ },
279
+ {
280
+ "epoch": 0.5233644859813084,
281
+ "grad_norm": 1.165434718132019,
282
+ "learning_rate": 8.823575730435693e-05,
283
+ "loss": 1.55217724,
284
+ "memory(GiB)": 32.07,
285
+ "step": 140,
286
+ "train_speed(iter/s)": 0.223153
287
+ },
288
+ {
289
+ "epoch": 0.5420560747663551,
290
+ "grad_norm": 1.1032906770706177,
291
+ "learning_rate": 8.721937040414481e-05,
292
+ "loss": 1.43740101,
293
+ "memory(GiB)": 32.07,
294
+ "step": 145,
295
+ "train_speed(iter/s)": 0.223845
296
+ },
297
+ {
298
+ "epoch": 0.5607476635514018,
299
+ "grad_norm": 1.1984739303588867,
300
+ "learning_rate": 8.616725966539832e-05,
301
+ "loss": 1.58604784,
302
+ "memory(GiB)": 32.07,
303
+ "step": 150,
304
+ "train_speed(iter/s)": 0.224618
305
+ },
306
+ {
307
+ "epoch": 0.5607476635514018,
308
+ "eval_loss": 1.4658682346343994,
309
+ "eval_runtime": 14.0407,
310
+ "eval_samples_per_second": 3.561,
311
+ "eval_steps_per_second": 3.561,
312
+ "step": 150
313
+ },
314
+ {
315
+ "epoch": 0.5794392523364486,
316
+ "grad_norm": 1.154517650604248,
317
+ "learning_rate": 8.508043492341944e-05,
318
+ "loss": 1.49082041,
319
+ "memory(GiB)": 32.07,
320
+ "step": 155,
321
+ "train_speed(iter/s)": 0.220462
322
+ },
323
+ {
324
+ "epoch": 0.5981308411214953,
325
+ "grad_norm": 1.2047632932662964,
326
+ "learning_rate": 8.395993933265101e-05,
327
+ "loss": 1.53753242,
328
+ "memory(GiB)": 32.07,
329
+ "step": 160,
330
+ "train_speed(iter/s)": 0.221167
331
+ },
332
+ {
333
+ "epoch": 0.616822429906542,
334
+ "grad_norm": 0.9952251315116882,
335
+ "learning_rate": 8.280684836543794e-05,
336
+ "loss": 1.49997816,
337
+ "memory(GiB)": 32.07,
338
+ "step": 165,
339
+ "train_speed(iter/s)": 0.22173
340
+ },
341
+ {
342
+ "epoch": 0.6355140186915887,
343
+ "grad_norm": 1.1730362176895142,
344
+ "learning_rate": 8.162226877976887e-05,
345
+ "loss": 1.50385504,
346
+ "memory(GiB)": 32.07,
347
+ "step": 170,
348
+ "train_speed(iter/s)": 0.222363
349
+ },
350
+ {
351
+ "epoch": 0.6542056074766355,
352
+ "grad_norm": 1.066243052482605,
353
+ "learning_rate": 8.040733755698955e-05,
354
+ "loss": 1.4824049,
355
+ "memory(GiB)": 32.07,
356
+ "step": 175,
357
+ "train_speed(iter/s)": 0.22299
358
+ },
359
+ {
360
+ "epoch": 0.6728971962616822,
361
+ "grad_norm": 1.2189449071884155,
362
+ "learning_rate": 7.916322081050709e-05,
363
+ "loss": 1.49032326,
364
+ "memory(GiB)": 32.07,
365
+ "step": 180,
366
+ "train_speed(iter/s)": 0.223605
367
+ },
368
+ {
369
+ "epoch": 0.6915887850467289,
370
+ "grad_norm": 1.07020103931427,
371
+ "learning_rate": 7.789111266653285e-05,
372
+ "loss": 1.46754303,
373
+ "memory(GiB)": 32.07,
374
+ "step": 185,
375
+ "train_speed(iter/s)": 0.224145
376
+ },
377
+ {
378
+ "epoch": 0.7102803738317757,
379
+ "grad_norm": 1.226481318473816,
380
+ "learning_rate": 7.659223411793798e-05,
381
+ "loss": 1.42194347,
382
+ "memory(GiB)": 32.07,
383
+ "step": 190,
384
+ "train_speed(iter/s)": 0.224687
385
+ },
386
+ {
387
+ "epoch": 0.7289719626168224,
388
+ "grad_norm": 1.111670732498169,
389
+ "learning_rate": 7.526783185232207e-05,
390
+ "loss": 1.50790215,
391
+ "memory(GiB)": 32.07,
392
+ "step": 195,
393
+ "train_speed(iter/s)": 0.225152
394
+ },
395
+ {
396
+ "epoch": 0.7476635514018691,
397
+ "grad_norm": 1.1171320676803589,
398
+ "learning_rate": 7.391917705541927e-05,
399
+ "loss": 1.51145458,
400
+ "memory(GiB)": 32.07,
401
+ "step": 200,
402
+ "train_speed(iter/s)": 0.22563
403
+ },
404
+ {
405
+ "epoch": 0.7476635514018691,
406
+ "eval_loss": 1.4480363130569458,
407
+ "eval_runtime": 14.0508,
408
+ "eval_samples_per_second": 3.559,
409
+ "eval_steps_per_second": 3.559,
410
+ "step": 200
411
+ },
412
+ {
413
+ "epoch": 0.7663551401869159,
414
+ "grad_norm": 0.9992289543151855,
415
+ "learning_rate": 7.254756419099074e-05,
416
+ "loss": 1.53672495,
417
+ "memory(GiB)": 32.07,
418
+ "step": 205,
419
+ "train_speed(iter/s)": 0.222373
420
+ },
421
+ {
422
+ "epoch": 0.7850467289719626,
423
+ "grad_norm": 1.076946496963501,
424
+ "learning_rate": 7.115430975837457e-05,
425
+ "loss": 1.51113377,
426
+ "memory(GiB)": 32.07,
427
+ "step": 210,
428
+ "train_speed(iter/s)": 0.222912
429
+ },
430
+ {
431
+ "epoch": 0.8037383177570093,
432
+ "grad_norm": 1.3144261837005615,
433
+ "learning_rate": 6.974075102888536e-05,
434
+ "loss": 1.51253147,
435
+ "memory(GiB)": 32.07,
436
+ "step": 215,
437
+ "train_speed(iter/s)": 0.223388
438
+ },
439
+ {
440
+ "epoch": 0.822429906542056,
441
+ "grad_norm": 1.2429286241531372,
442
+ "learning_rate": 6.830824476227646e-05,
443
+ "loss": 1.49584999,
444
+ "memory(GiB)": 32.07,
445
+ "step": 220,
446
+ "train_speed(iter/s)": 0.22384
447
+ },
448
+ {
449
+ "epoch": 0.8411214953271028,
450
+ "grad_norm": 1.213188886642456,
451
+ "learning_rate": 6.685816590449708e-05,
452
+ "loss": 1.4517292,
453
+ "memory(GiB)": 32.07,
454
+ "step": 225,
455
+ "train_speed(iter/s)": 0.224262
456
+ },
457
+ {
458
+ "epoch": 0.8598130841121495,
459
+ "grad_norm": 1.1008031368255615,
460
+ "learning_rate": 6.539190626799366e-05,
461
+ "loss": 1.44860907,
462
+ "memory(GiB)": 32.07,
463
+ "step": 230,
464
+ "train_speed(iter/s)": 0.224691
465
+ },
466
+ {
467
+ "epoch": 0.8785046728971962,
468
+ "grad_norm": 1.105083703994751,
469
+ "learning_rate": 6.391087319582264e-05,
470
+ "loss": 1.45654058,
471
+ "memory(GiB)": 32.07,
472
+ "step": 235,
473
+ "train_speed(iter/s)": 0.225105
474
+ },
475
+ {
476
+ "epoch": 0.897196261682243,
477
+ "grad_norm": 1.1485651731491089,
478
+ "learning_rate": 6.241648821085666e-05,
479
+ "loss": 1.4626853,
480
+ "memory(GiB)": 32.07,
481
+ "step": 240,
482
+ "train_speed(iter/s)": 0.225456
483
+ },
484
+ {
485
+ "epoch": 0.9158878504672897,
486
+ "grad_norm": 1.2288539409637451,
487
+ "learning_rate": 6.0910185651380626e-05,
488
+ "loss": 1.41080866,
489
+ "memory(GiB)": 32.07,
490
+ "step": 245,
491
+ "train_speed(iter/s)": 0.225881
492
+ },
493
+ {
494
+ "epoch": 0.9345794392523364,
495
+ "grad_norm": 1.2186890840530396,
496
+ "learning_rate": 5.939341129438739e-05,
497
+ "loss": 1.53512402,
498
+ "memory(GiB)": 32.07,
499
+ "step": 250,
500
+ "train_speed(iter/s)": 0.226215
501
+ },
502
+ {
503
+ "epoch": 0.9345794392523364,
504
+ "eval_loss": 1.438408374786377,
505
+ "eval_runtime": 14.1598,
506
+ "eval_samples_per_second": 3.531,
507
+ "eval_steps_per_second": 3.531,
508
+ "step": 250
509
+ },
510
+ {
511
+ "epoch": 0.9532710280373832,
512
+ "grad_norm": 1.1940230131149292,
513
+ "learning_rate": 5.786762096789431e-05,
514
+ "loss": 1.55513544,
515
+ "memory(GiB)": 32.07,
516
+ "step": 255,
517
+ "train_speed(iter/s)": 0.223566
518
+ },
519
+ {
520
+ "epoch": 0.9719626168224299,
521
+ "grad_norm": 1.0835857391357422,
522
+ "learning_rate": 5.633427915361261e-05,
523
+ "loss": 1.51988029,
524
+ "memory(GiB)": 32.07,
525
+ "step": 260,
526
+ "train_speed(iter/s)": 0.22394
527
+ },
528
+ {
529
+ "epoch": 0.9906542056074766,
530
+ "grad_norm": 1.170660376548767,
531
+ "learning_rate": 5.479485758131089e-05,
532
+ "loss": 1.56143446,
533
+ "memory(GiB)": 32.07,
534
+ "step": 265,
535
+ "train_speed(iter/s)": 0.224337
536
+ },
537
+ {
538
+ "epoch": 1.0093457943925233,
539
+ "grad_norm": 1.0278513431549072,
540
+ "learning_rate": 5.325083381622165e-05,
541
+ "loss": 1.43758631,
542
+ "memory(GiB)": 32.07,
543
+ "step": 270,
544
+ "train_speed(iter/s)": 0.224666
545
+ },
546
+ {
547
+ "epoch": 1.02803738317757,
548
+ "grad_norm": 1.13231360912323,
549
+ "learning_rate": 5.1703689840846945e-05,
550
+ "loss": 1.34864044,
551
+ "memory(GiB)": 32.07,
552
+ "step": 275,
553
+ "train_speed(iter/s)": 0.225034
554
+ },
555
+ {
556
+ "epoch": 1.0467289719626167,
557
+ "grad_norm": 1.2419425249099731,
558
+ "learning_rate": 5.01549106325243e-05,
559
+ "loss": 1.38481417,
560
+ "memory(GiB)": 32.07,
561
+ "step": 280,
562
+ "train_speed(iter/s)": 0.225399
563
+ },
564
+ {
565
+ "epoch": 1.0654205607476634,
566
+ "grad_norm": 1.336288332939148,
567
+ "learning_rate": 4.860598273811792e-05,
568
+ "loss": 1.24492655,
569
+ "memory(GiB)": 32.07,
570
+ "step": 285,
571
+ "train_speed(iter/s)": 0.225789
572
+ },
573
+ {
574
+ "epoch": 1.0841121495327102,
575
+ "grad_norm": 1.241809368133545,
576
+ "learning_rate": 4.705839284720376e-05,
577
+ "loss": 1.36301146,
578
+ "memory(GiB)": 32.07,
579
+ "step": 290,
580
+ "train_speed(iter/s)": 0.226105
581
+ },
582
+ {
583
+ "epoch": 1.102803738317757,
584
+ "grad_norm": 1.4412420988082886,
585
+ "learning_rate": 4.55136263651172e-05,
586
+ "loss": 1.39876356,
587
+ "memory(GiB)": 32.07,
588
+ "step": 295,
589
+ "train_speed(iter/s)": 0.226405
590
+ },
591
+ {
592
+ "epoch": 1.1214953271028036,
593
+ "grad_norm": 1.6165404319763184,
594
+ "learning_rate": 4.397316598723385e-05,
595
+ "loss": 1.32808571,
596
+ "memory(GiB)": 32.07,
597
+ "step": 300,
598
+ "train_speed(iter/s)": 0.226709
599
+ },
600
+ {
601
+ "epoch": 1.1214953271028036,
602
+ "eval_loss": 1.4294430017471313,
603
+ "eval_runtime": 14.1178,
604
+ "eval_samples_per_second": 3.542,
605
+ "eval_steps_per_second": 3.542,
606
+ "step": 300
607
+ },
608
+ {
609
+ "epoch": 1.1401869158878504,
610
+ "grad_norm": 1.4734883308410645,
611
+ "learning_rate": 4.243849027585096e-05,
612
+ "loss": 1.37022314,
613
+ "memory(GiB)": 32.07,
614
+ "step": 305,
615
+ "train_speed(iter/s)": 0.224508
616
+ },
617
+ {
618
+ "epoch": 1.158878504672897,
619
+ "grad_norm": 1.5161515474319458,
620
+ "learning_rate": 4.0911072241036194e-05,
621
+ "loss": 1.40692539,
622
+ "memory(GiB)": 32.07,
623
+ "step": 310,
624
+ "train_speed(iter/s)": 0.224822
625
+ },
626
+ {
627
+ "epoch": 1.1775700934579438,
628
+ "grad_norm": 1.4354695081710815,
629
+ "learning_rate": 3.9392377926805226e-05,
630
+ "loss": 1.31709337,
631
+ "memory(GiB)": 32.07,
632
+ "step": 315,
633
+ "train_speed(iter/s)": 0.225147
634
+ },
635
+ {
636
+ "epoch": 1.1962616822429906,
637
+ "grad_norm": 1.5612841844558716,
638
+ "learning_rate": 3.788386500398583e-05,
639
+ "loss": 1.38046598,
640
+ "memory(GiB)": 32.07,
641
+ "step": 320,
642
+ "train_speed(iter/s)": 0.225425
643
+ },
644
+ {
645
+ "epoch": 1.2149532710280373,
646
+ "grad_norm": 1.353385090827942,
647
+ "learning_rate": 3.6386981371118355e-05,
648
+ "loss": 1.29831305,
649
+ "memory(GiB)": 32.07,
650
+ "step": 325,
651
+ "train_speed(iter/s)": 0.225693
652
+ },
653
+ {
654
+ "epoch": 1.233644859813084,
655
+ "grad_norm": 1.6214525699615479,
656
+ "learning_rate": 3.49031637647361e-05,
657
+ "loss": 1.33498459,
658
+ "memory(GiB)": 32.07,
659
+ "step": 330,
660
+ "train_speed(iter/s)": 0.225991
661
+ },
662
+ {
663
+ "epoch": 1.2523364485981308,
664
+ "grad_norm": 1.441267490386963,
665
+ "learning_rate": 3.343383638035902e-05,
666
+ "loss": 1.2935997,
667
+ "memory(GiB)": 32.07,
668
+ "step": 335,
669
+ "train_speed(iter/s)": 0.226241
670
+ },
671
+ {
672
+ "epoch": 1.2710280373831775,
673
+ "grad_norm": 1.5621421337127686,
674
+ "learning_rate": 3.1980409505524544e-05,
675
+ "loss": 1.32472296,
676
+ "memory(GiB)": 32.07,
677
+ "step": 340,
678
+ "train_speed(iter/s)": 0.226507
679
+ },
680
+ {
681
+ "epoch": 1.2897196261682242,
682
+ "grad_norm": 1.7050727605819702,
683
+ "learning_rate": 3.054427816616773e-05,
684
+ "loss": 1.25045223,
685
+ "memory(GiB)": 32.07,
686
+ "step": 345,
687
+ "train_speed(iter/s)": 0.226762
688
+ },
689
+ {
690
+ "epoch": 1.308411214953271,
691
+ "grad_norm": 1.5206207036972046,
692
+ "learning_rate": 2.91268207876494e-05,
693
+ "loss": 1.33886337,
694
+ "memory(GiB)": 32.07,
695
+ "step": 350,
696
+ "train_speed(iter/s)": 0.226984
697
+ },
698
+ {
699
+ "epoch": 1.308411214953271,
700
+ "eval_loss": 1.4251823425292969,
701
+ "eval_runtime": 14.0593,
702
+ "eval_samples_per_second": 3.556,
703
+ "eval_steps_per_second": 3.556,
704
+ "step": 350
705
+ },
706
+ {
707
+ "epoch": 1.3271028037383177,
708
+ "grad_norm": 1.5205532312393188,
709
+ "learning_rate": 2.7729397871718304e-05,
710
+ "loss": 1.28512764,
711
+ "memory(GiB)": 32.07,
712
+ "step": 355,
713
+ "train_speed(iter/s)": 0.225063
714
+ },
715
+ {
716
+ "epoch": 1.3457943925233644,
717
+ "grad_norm": 1.5533926486968994,
718
+ "learning_rate": 2.635335069067617e-05,
719
+ "loss": 1.30997047,
720
+ "memory(GiB)": 32.07,
721
+ "step": 360,
722
+ "train_speed(iter/s)": 0.225339
723
+ },
724
+ {
725
+ "epoch": 1.3644859813084111,
726
+ "grad_norm": 1.581883192062378,
727
+ "learning_rate": 2.500000000000001e-05,
728
+ "loss": 1.28296366,
729
+ "memory(GiB)": 32.07,
730
+ "step": 365,
731
+ "train_speed(iter/s)": 0.225599
732
+ },
733
+ {
734
+ "epoch": 1.3831775700934579,
735
+ "grad_norm": 1.4634901285171509,
736
+ "learning_rate": 2.367064477065652e-05,
737
+ "loss": 1.31434088,
738
+ "memory(GiB)": 32.07,
739
+ "step": 370,
740
+ "train_speed(iter/s)": 0.225861
741
+ },
742
+ {
743
+ "epoch": 1.4018691588785046,
744
+ "grad_norm": 1.706288456916809,
745
+ "learning_rate": 2.2366560942325832e-05,
746
+ "loss": 1.30933113,
747
+ "memory(GiB)": 32.07,
748
+ "step": 375,
749
+ "train_speed(iter/s)": 0.226141
750
+ },
751
+ {
752
+ "epoch": 1.4205607476635513,
753
+ "grad_norm": 1.6696898937225342,
754
+ "learning_rate": 2.108900019873103e-05,
755
+ "loss": 1.32816324,
756
+ "memory(GiB)": 32.07,
757
+ "step": 380,
758
+ "train_speed(iter/s)": 0.226392
759
+ },
760
+ {
761
+ "epoch": 1.439252336448598,
762
+ "grad_norm": 1.590394377708435,
763
+ "learning_rate": 1.983918876624902e-05,
764
+ "loss": 1.26775227,
765
+ "memory(GiB)": 32.07,
766
+ "step": 385,
767
+ "train_speed(iter/s)": 0.226637
768
+ },
769
+ {
770
+ "epoch": 1.4579439252336448,
771
+ "grad_norm": 1.7391793727874756,
772
+ "learning_rate": 1.8618326236955907e-05,
773
+ "loss": 1.33946781,
774
+ "memory(GiB)": 32.07,
775
+ "step": 390,
776
+ "train_speed(iter/s)": 0.226874
777
+ },
778
+ {
779
+ "epoch": 1.4766355140186915,
780
+ "grad_norm": 1.7035928964614868,
781
+ "learning_rate": 1.7427584417236194e-05,
782
+ "loss": 1.34862604,
783
+ "memory(GiB)": 32.07,
784
+ "step": 395,
785
+ "train_speed(iter/s)": 0.227069
786
+ },
787
+ {
788
+ "epoch": 1.4953271028037383,
789
+ "grad_norm": 1.5830693244934082,
790
+ "learning_rate": 1.626810620306163e-05,
791
+ "loss": 1.27288446,
792
+ "memory(GiB)": 32.07,
793
+ "step": 400,
794
+ "train_speed(iter/s)": 0.227266
795
+ },
796
+ {
797
+ "epoch": 1.4953271028037383,
798
+ "eval_loss": 1.4195191860198975,
799
+ "eval_runtime": 14.0879,
800
+ "eval_samples_per_second": 3.549,
801
+ "eval_steps_per_second": 3.549,
802
+ "step": 400
803
+ },
804
+ {
805
+ "epoch": 1.514018691588785,
806
+ "grad_norm": 1.4809561967849731,
807
+ "learning_rate": 1.5141004483018323e-05,
808
+ "loss": 1.31938076,
809
+ "memory(GiB)": 32.07,
810
+ "step": 405,
811
+ "train_speed(iter/s)": 0.225559
812
+ },
813
+ {
814
+ "epoch": 1.5327102803738317,
815
+ "grad_norm": 1.8556567430496216,
816
+ "learning_rate": 1.4047361070135995e-05,
817
+ "loss": 1.33600292,
818
+ "memory(GiB)": 32.07,
819
+ "step": 410,
820
+ "train_speed(iter/s)": 0.225804
821
+ },
822
+ {
823
+ "epoch": 1.5514018691588785,
824
+ "grad_norm": 1.5470691919326782,
825
+ "learning_rate": 1.2988225663543602e-05,
826
+ "loss": 1.40292425,
827
+ "memory(GiB)": 32.07,
828
+ "step": 415,
829
+ "train_speed(iter/s)": 0.226027
830
+ },
831
+ {
832
+ "epoch": 1.5700934579439252,
833
+ "grad_norm": 1.8364381790161133,
834
+ "learning_rate": 1.1964614840949002e-05,
835
+ "loss": 1.32833939,
836
+ "memory(GiB)": 32.07,
837
+ "step": 420,
838
+ "train_speed(iter/s)": 0.226267
839
+ },
840
+ {
841
+ "epoch": 1.588785046728972,
842
+ "grad_norm": 1.6938135623931885,
843
+ "learning_rate": 1.097751108290867e-05,
844
+ "loss": 1.35209036,
845
+ "memory(GiB)": 32.07,
846
+ "step": 425,
847
+ "train_speed(iter/s)": 0.226474
848
+ },
849
+ {
850
+ "epoch": 1.6074766355140186,
851
+ "grad_norm": 1.7861816883087158,
852
+ "learning_rate": 1.0027861829824952e-05,
853
+ "loss": 1.27312994,
854
+ "memory(GiB)": 32.07,
855
+ "step": 430,
856
+ "train_speed(iter/s)": 0.226695
857
+ },
858
+ {
859
+ "epoch": 1.6261682242990654,
860
+ "grad_norm": 1.6619056463241577,
861
+ "learning_rate": 9.11657857257509e-06,
862
+ "loss": 1.35062437,
863
+ "memory(GiB)": 32.07,
864
+ "step": 435,
865
+ "train_speed(iter/s)": 0.22691
866
+ },
867
+ {
868
+ "epoch": 1.644859813084112,
869
+ "grad_norm": 1.7696343660354614,
870
+ "learning_rate": 8.244535977645585e-06,
871
+ "loss": 1.32785254,
872
+ "memory(GiB)": 32.07,
873
+ "step": 440,
874
+ "train_speed(iter/s)": 0.227108
875
+ },
876
+ {
877
+ "epoch": 1.6635514018691588,
878
+ "grad_norm": 1.6938729286193848,
879
+ "learning_rate": 7.412571047611155e-06,
880
+ "loss": 1.3087183,
881
+ "memory(GiB)": 32.07,
882
+ "step": 445,
883
+ "train_speed(iter/s)": 0.227305
884
+ },
885
+ {
886
+ "epoch": 1.6822429906542056,
887
+ "grad_norm": 1.7258585691452026,
888
+ "learning_rate": 6.621482317764105e-06,
889
+ "loss": 1.30971994,
890
+ "memory(GiB)": 32.07,
891
+ "step": 450,
892
+ "train_speed(iter/s)": 0.227507
893
+ },
894
+ {
895
+ "epoch": 1.6822429906542056,
896
+ "eval_loss": 1.4193787574768066,
897
+ "eval_runtime": 17.3944,
898
+ "eval_samples_per_second": 2.874,
899
+ "eval_steps_per_second": 2.874,
900
+ "step": 450
901
+ },
902
+ {
903
+ "epoch": 1.7009345794392523,
904
+ "grad_norm": 1.8556472063064575,
905
+ "learning_rate": 5.872029089665587e-06,
906
+ "loss": 1.26630516,
907
+ "memory(GiB)": 32.07,
908
+ "step": 455,
909
+ "train_speed(iter/s)": 0.225602
910
+ },
911
+ {
912
+ "epoch": 1.719626168224299,
913
+ "grad_norm": 1.852525234222412,
914
+ "learning_rate": 5.164930702353782e-06,
915
+ "loss": 1.34138193,
916
+ "memory(GiB)": 32.07,
917
+ "step": 460,
918
+ "train_speed(iter/s)": 0.225826
919
+ },
920
+ {
921
+ "epoch": 1.7383177570093458,
922
+ "grad_norm": 1.557905673980713,
923
+ "learning_rate": 4.500865841909168e-06,
924
+ "loss": 1.30747194,
925
+ "memory(GiB)": 32.07,
926
+ "step": 465,
927
+ "train_speed(iter/s)": 0.225996
928
+ },
929
+ {
930
+ "epoch": 1.7570093457943925,
931
+ "grad_norm": 1.7627642154693604,
932
+ "learning_rate": 3.880471890038967e-06,
933
+ "loss": 1.34135695,
934
+ "memory(GiB)": 32.07,
935
+ "step": 470,
936
+ "train_speed(iter/s)": 0.226199
937
+ },
938
+ {
939
+ "epoch": 1.7757009345794392,
940
+ "grad_norm": 1.4336940050125122,
941
+ "learning_rate": 3.3043443123065286e-06,
942
+ "loss": 1.38070517,
943
+ "memory(GiB)": 32.07,
944
+ "step": 475,
945
+ "train_speed(iter/s)": 0.226358
946
+ },
947
+ {
948
+ "epoch": 1.794392523364486,
949
+ "grad_norm": 1.906886339187622,
950
+ "learning_rate": 2.7730360865923956e-06,
951
+ "loss": 1.34674683,
952
+ "memory(GiB)": 32.07,
953
+ "step": 480,
954
+ "train_speed(iter/s)": 0.226536
955
+ },
956
+ {
957
+ "epoch": 1.8130841121495327,
958
+ "grad_norm": 1.7454955577850342,
959
+ "learning_rate": 2.287057172336021e-06,
960
+ "loss": 1.38749065,
961
+ "memory(GiB)": 32.07,
962
+ "step": 485,
963
+ "train_speed(iter/s)": 0.226699
964
+ },
965
+ {
966
+ "epoch": 1.8317757009345794,
967
+ "grad_norm": 1.7366608381271362,
968
+ "learning_rate": 1.8468740210672076e-06,
969
+ "loss": 1.30795374,
970
+ "memory(GiB)": 32.07,
971
+ "step": 490,
972
+ "train_speed(iter/s)": 0.226882
973
+ },
974
+ {
975
+ "epoch": 1.8504672897196262,
976
+ "grad_norm": 1.5829346179962158,
977
+ "learning_rate": 1.4529091286973995e-06,
978
+ "loss": 1.32902784,
979
+ "memory(GiB)": 32.07,
980
+ "step": 495,
981
+ "train_speed(iter/s)": 0.227039
982
+ },
983
+ {
984
+ "epoch": 1.8691588785046729,
985
+ "grad_norm": 1.6908546686172485,
986
+ "learning_rate": 1.1055406300002347e-06,
987
+ "loss": 1.33979492,
988
+ "memory(GiB)": 32.07,
989
+ "step": 500,
990
+ "train_speed(iter/s)": 0.227213
991
+ },
992
+ {
993
+ "epoch": 1.8691588785046729,
994
+ "eval_loss": 1.4183509349822998,
995
+ "eval_runtime": 14.1519,
996
+ "eval_samples_per_second": 3.533,
997
+ "eval_steps_per_second": 3.533,
998
+ "step": 500
999
+ },
1000
+ {
1001
+ "epoch": 1.8878504672897196,
1002
+ "grad_norm": 1.915726900100708,
1003
+ "learning_rate": 8.0510193567086e-07,
1004
+ "loss": 1.30009985,
1005
+ "memory(GiB)": 32.07,
1006
+ "step": 505,
1007
+ "train_speed(iter/s)": 0.225856
1008
+ },
1009
+ {
1010
+ "epoch": 1.9065420560747663,
1011
+ "grad_norm": 1.6646161079406738,
1012
+ "learning_rate": 5.518814123121885e-07,
1013
+ "loss": 1.37087755,
1014
+ "memory(GiB)": 32.07,
1015
+ "step": 510,
1016
+ "train_speed(iter/s)": 0.226034
1017
+ },
1018
+ {
1019
+ "epoch": 1.925233644859813,
1020
+ "grad_norm": 1.7108522653579712,
1021
+ "learning_rate": 3.4612210565528326e-07,
1022
+ "loss": 1.35631628,
1023
+ "memory(GiB)": 32.07,
1024
+ "step": 515,
1025
+ "train_speed(iter/s)": 0.22621
1026
+ },
1027
+ {
1028
+ "epoch": 1.9439252336448598,
1029
+ "grad_norm": 1.7579667568206787,
1030
+ "learning_rate": 1.8802150727962876e-07,
1031
+ "loss": 1.24607553,
1032
+ "memory(GiB)": 32.07,
1033
+ "step": 520,
1034
+ "train_speed(iter/s)": 0.226384
1035
+ },
1036
+ {
1037
+ "epoch": 1.9626168224299065,
1038
+ "grad_norm": 1.634746789932251,
1039
+ "learning_rate": 7.773136505700995e-08,
1040
+ "loss": 1.27467356,
1041
+ "memory(GiB)": 32.07,
1042
+ "step": 525,
1043
+ "train_speed(iter/s)": 0.226543
1044
+ },
1045
+ {
1046
+ "epoch": 1.9813084112149533,
1047
+ "grad_norm": 1.620557188987732,
1048
+ "learning_rate": 1.5357537501159423e-08,
1049
+ "loss": 1.318472,
1050
+ "memory(GiB)": 32.07,
1051
+ "step": 530,
1052
+ "train_speed(iter/s)": 0.226703
1053
+ },
1054
+ {
1055
+ "epoch": 1.9962616822429906,
1056
+ "eval_loss": 1.4178756475448608,
1057
+ "eval_runtime": 14.1624,
1058
+ "eval_samples_per_second": 3.53,
1059
+ "eval_steps_per_second": 3.53,
1060
+ "step": 534
1061
+ }
1062
+ ],
1063
+ "logging_steps": 5,
1064
+ "max_steps": 534,
1065
+ "num_input_tokens_seen": 0,
1066
+ "num_train_epochs": 2,
1067
+ "save_steps": 50,
1068
+ "stateful_callbacks": {
1069
+ "TrainerControl": {
1070
+ "args": {
1071
+ "should_epoch_stop": false,
1072
+ "should_evaluate": false,
1073
+ "should_log": false,
1074
+ "should_save": true,
1075
+ "should_training_stop": true
1076
+ },
1077
+ "attributes": {}
1078
+ }
1079
+ },
1080
+ "total_flos": 2.73270917085696e+16,
1081
+ "train_batch_size": 1,
1082
+ "trial_name": null,
1083
+ "trial_params": null
1084
+ }
llava1_6-vicuna-7b-instruct/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f1ee500bc16818ccb6af6404742071ed8aa5c214d899905af5c3d73f16eae7b
3
+ size 7416
minicpm-v-v2_6-chat/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: /fs/clip-projects/geoguesser/vlms/cpm/MiniCPM-V-2_6
3
+ library_name: peft
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+ ### Framework versions
201
+
202
+ - PEFT 0.12.0
minicpm-v-v2_6-chat/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "/fs/clip-projects/geoguesser/vlms/cpm/MiniCPM-V-2_6",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": [],
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "^(llm|resampler)(?!.*(lm_head|output|emb|wte|shared)).*",
23
+ "task_type": "CAUSAL_LM",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
minicpm-v-v2_6-chat/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba0c23a6f4291a7bf0ac08aadc2782defa427169e319594f0cc95f379f5c2a6f
3
+ size 81175080
minicpm-v-v2_6-chat/additional_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}
minicpm-v-v2_6-chat/configuration.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "framework": "Pytorch",
3
+ "task": "visual-question-answering",
4
+ "adapter_cfg": {
5
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/cpm/MiniCPM-V-2_6",
6
+ "model_revision": "master",
7
+ "sft_type": "lora",
8
+ "tuner_backend": "peft",
9
+ "template_type": "minicpm-v-v2_6",
10
+ "dtype": "bf16",
11
+ "system": "You are a helpful assistant."
12
+ }
13
+ }
minicpm-v-v2_6-chat/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "eos_token_id": 151645,
4
+ "max_new_tokens": 2048,
5
+ "pad_token_id": 151643,
6
+ "transformers_version": "4.45.1"
7
+ }
minicpm-v-v2_6-chat/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f804033311c05e15514241ec05e2702e77e1ea22c7ddc34080f9960abca0dd0
3
+ size 162575542
minicpm-v-v2_6-chat/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c267f7c126c2e718bba599dcd30e9c477a5b3cf23ec7812edda9d522e67d4a1b
3
+ size 14244
minicpm-v-v2_6-chat/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b75bb0cab5c1ac64f1f42e99abb3e1c6d095394f6dd0c73c81aee621b476d6d9
3
+ size 1064
minicpm-v-v2_6-chat/sft_args.json ADDED
@@ -0,0 +1,247 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "minicpm-v-v2_6-chat",
3
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/cpm/MiniCPM-V-2_6",
4
+ "model_revision": "master",
5
+ "full_determinism": false,
6
+ "sft_type": "lora",
7
+ "freeze_parameters": [],
8
+ "freeze_vit": false,
9
+ "freeze_parameters_ratio": 0.0,
10
+ "additional_trainable_parameters": [],
11
+ "tuner_backend": "peft",
12
+ "template_type": "minicpm-v-v2_6",
13
+ "output_dir": "/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955",
14
+ "add_output_dir_suffix": true,
15
+ "ddp_backend": null,
16
+ "ddp_find_unused_parameters": null,
17
+ "ddp_broadcast_buffers": null,
18
+ "ddp_timeout": 1800,
19
+ "seed": 42,
20
+ "resume_from_checkpoint": null,
21
+ "resume_only_model": false,
22
+ "ignore_data_skip": false,
23
+ "dtype": "bf16",
24
+ "packing": false,
25
+ "train_backend": "transformers",
26
+ "tp": 1,
27
+ "pp": 1,
28
+ "min_lr": null,
29
+ "sequence_parallel": false,
30
+ "model_kwargs": null,
31
+ "loss_name": null,
32
+ "dataset": [
33
+ "train_set_a.jsonl"
34
+ ],
35
+ "val_dataset": [
36
+ "test_set_a.jsonl"
37
+ ],
38
+ "dataset_seed": 42,
39
+ "dataset_test_ratio": 0.0,
40
+ "use_loss_scale": false,
41
+ "loss_scale_config_path": "/fs/clip-projects/geoguesser/zheyuan/DPO/swift/swift/llm/agent/default_loss_scale_config.json",
42
+ "system": "You are a helpful assistant.",
43
+ "tools_prompt": "react_en",
44
+ "max_length": 2048,
45
+ "truncation_strategy": "delete",
46
+ "check_dataset_strategy": "none",
47
+ "streaming": false,
48
+ "streaming_val_size": 0,
49
+ "streaming_buffer_size": 16384,
50
+ "model_name": [
51
+ null,
52
+ null
53
+ ],
54
+ "model_author": [
55
+ null,
56
+ null
57
+ ],
58
+ "quant_method": null,
59
+ "quantization_bit": 0,
60
+ "hqq_axis": 0,
61
+ "hqq_dynamic_config_path": null,
62
+ "bnb_4bit_comp_dtype": "bf16",
63
+ "bnb_4bit_quant_type": "nf4",
64
+ "bnb_4bit_use_double_quant": true,
65
+ "bnb_4bit_quant_storage": null,
66
+ "rescale_image": -1,
67
+ "target_modules": "^(llm|resampler)(?!.*(lm_head|output|emb|wte|shared)).*",
68
+ "target_regex": null,
69
+ "modules_to_save": [],
70
+ "lora_rank": 8,
71
+ "lora_alpha": 32,
72
+ "lora_dropout": 0.05,
73
+ "lora_bias_trainable": "none",
74
+ "lora_dtype": null,
75
+ "lora_lr_ratio": null,
76
+ "use_rslora": false,
77
+ "use_dora": false,
78
+ "init_lora_weights": true,
79
+ "fourier_n_frequency": 2000,
80
+ "fourier_scaling": 300.0,
81
+ "rope_scaling": null,
82
+ "boft_block_size": 4,
83
+ "boft_block_num": 0,
84
+ "boft_n_butterfly_factor": 1,
85
+ "boft_dropout": 0.0,
86
+ "vera_rank": 256,
87
+ "vera_projection_prng_key": 0,
88
+ "vera_dropout": 0.0,
89
+ "vera_d_initial": 0.1,
90
+ "adapter_act": "gelu",
91
+ "adapter_length": 128,
92
+ "use_galore": false,
93
+ "galore_target_modules": null,
94
+ "galore_rank": 128,
95
+ "galore_update_proj_gap": 50,
96
+ "galore_scale": 1.0,
97
+ "galore_proj_type": "std",
98
+ "galore_optim_per_parameter": false,
99
+ "galore_with_embedding": false,
100
+ "galore_quantization": false,
101
+ "galore_proj_quant": false,
102
+ "galore_proj_bits": 4,
103
+ "galore_proj_group_size": 256,
104
+ "galore_cos_threshold": 0.4,
105
+ "galore_gamma_proj": 2,
106
+ "galore_queue_size": 5,
107
+ "adalora_target_r": 8,
108
+ "adalora_init_r": 12,
109
+ "adalora_tinit": 0,
110
+ "adalora_tfinal": 0,
111
+ "adalora_deltaT": 1,
112
+ "adalora_beta1": 0.85,
113
+ "adalora_beta2": 0.85,
114
+ "adalora_orth_reg_weight": 0.5,
115
+ "ia3_feedforward_modules": [],
116
+ "llamapro_num_new_blocks": 4,
117
+ "llamapro_num_groups": null,
118
+ "neftune_noise_alpha": null,
119
+ "neftune_backend": "transformers",
120
+ "lisa_activated_layers": 0,
121
+ "lisa_step_interval": 20,
122
+ "reft_layer_key": null,
123
+ "reft_layers": null,
124
+ "reft_rank": 4,
125
+ "reft_intervention_type": "LoreftIntervention",
126
+ "reft_args": null,
127
+ "use_liger": false,
128
+ "gradient_checkpointing": true,
129
+ "deepspeed": null,
130
+ "batch_size": 1,
131
+ "eval_batch_size": 1,
132
+ "auto_find_batch_size": false,
133
+ "num_train_epochs": 2,
134
+ "max_steps": -1,
135
+ "optim": "adamw_torch",
136
+ "adam_beta1": 0.9,
137
+ "adam_beta2": 0.95,
138
+ "adam_epsilon": 1e-08,
139
+ "learning_rate": 0.0001,
140
+ "weight_decay": 0.1,
141
+ "gradient_accumulation_steps": 4,
142
+ "max_grad_norm": 1,
143
+ "predict_with_generate": false,
144
+ "lr_scheduler_type": "cosine",
145
+ "lr_scheduler_kwargs": {},
146
+ "warmup_ratio": 0.05,
147
+ "warmup_steps": 0,
148
+ "eval_steps": 50,
149
+ "save_steps": 50,
150
+ "save_only_model": false,
151
+ "save_total_limit": 2,
152
+ "logging_steps": 5,
153
+ "acc_steps": 1,
154
+ "dataloader_num_workers": 1,
155
+ "dataloader_pin_memory": true,
156
+ "dataloader_drop_last": false,
157
+ "push_to_hub": false,
158
+ "hub_model_id": null,
159
+ "hub_token": null,
160
+ "hub_private_repo": false,
161
+ "hub_strategy": "every_save",
162
+ "test_oom_error": false,
163
+ "disable_tqdm": false,
164
+ "lazy_tokenize": true,
165
+ "preprocess_num_proc": 1,
166
+ "use_flash_attn": null,
167
+ "ignore_args_error": false,
168
+ "check_model_is_latest": true,
169
+ "logging_dir": "/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955/runs",
170
+ "report_to": [
171
+ "tensorboard"
172
+ ],
173
+ "acc_strategy": "token",
174
+ "save_on_each_node": false,
175
+ "evaluation_strategy": "steps",
176
+ "save_strategy": "steps",
177
+ "save_safetensors": true,
178
+ "gpu_memory_fraction": null,
179
+ "include_num_input_tokens_seen": false,
180
+ "local_repo_path": null,
181
+ "custom_register_path": null,
182
+ "custom_dataset_info": null,
183
+ "device_map_config": null,
184
+ "device_max_memory": [],
185
+ "max_new_tokens": 2048,
186
+ "do_sample": null,
187
+ "temperature": null,
188
+ "top_k": null,
189
+ "top_p": null,
190
+ "repetition_penalty": null,
191
+ "num_beams": 1,
192
+ "fsdp": "",
193
+ "fsdp_config": null,
194
+ "sequence_parallel_size": 1,
195
+ "model_layer_cls_name": null,
196
+ "metric_warmup_step": 0,
197
+ "fsdp_num": 1,
198
+ "per_device_train_batch_size": null,
199
+ "per_device_eval_batch_size": null,
200
+ "eval_strategy": null,
201
+ "self_cognition_sample": 0,
202
+ "train_dataset_mix_ratio": 0.0,
203
+ "train_dataset_mix_ds": [
204
+ "ms-bench"
205
+ ],
206
+ "train_dataset_sample": -1,
207
+ "val_dataset_sample": null,
208
+ "safe_serialization": null,
209
+ "only_save_model": null,
210
+ "neftune_alpha": null,
211
+ "deepspeed_config_path": null,
212
+ "model_cache_dir": null,
213
+ "lora_dropout_p": null,
214
+ "lora_target_modules": [],
215
+ "lora_target_regex": null,
216
+ "lora_modules_to_save": [],
217
+ "boft_target_modules": [],
218
+ "boft_modules_to_save": [],
219
+ "vera_target_modules": [],
220
+ "vera_modules_to_save": [],
221
+ "ia3_target_modules": [],
222
+ "ia3_modules_to_save": [],
223
+ "custom_train_dataset_path": [],
224
+ "custom_val_dataset_path": [],
225
+ "device_map_config_path": null,
226
+ "push_hub_strategy": null,
227
+ "use_self_cognition": false,
228
+ "is_multimodal": true,
229
+ "is_vision": true,
230
+ "lora_use_embedding": false,
231
+ "lora_use_all": false,
232
+ "lora_m2s_use_embedding": false,
233
+ "lora_m2s_use_ln": false,
234
+ "torch_dtype": "torch.bfloat16",
235
+ "fp16": false,
236
+ "bf16": true,
237
+ "rank": -1,
238
+ "local_rank": -1,
239
+ "world_size": 1,
240
+ "local_world_size": 1,
241
+ "bnb_4bit_compute_dtype": "torch.bfloat16",
242
+ "load_in_4bit": false,
243
+ "load_in_8bit": false,
244
+ "train_sampler_random": true,
245
+ "train_type": "sft",
246
+ "training_args": "Seq2SeqTrainingArguments(output_dir='/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=4, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0001, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1, num_train_epochs=2, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs={}, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=5, logging_nan_inf_filter=True, save_strategy=<IntervalStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=2, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=50, dataloader_num_workers=1, dataloader_prefetch_factor=None, past_index=-1, run_name='/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed=None, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=False, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy=None, push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=False, include_num_input_tokens_seen=False, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=GenerationConfig {\n \"bos_token_id\": 151643,\n \"eos_token_id\": 151645,\n \"max_new_tokens\": 2048,\n \"pad_token_id\": 151643\n}\n, acc_strategy='token', loss_name=None, additional_saved_files=[], train_sampler_random=True, metric_warmup_step=0, train_dataset_sample=-1)"
247
+ }
minicpm-v-v2_6-chat/trainer_state.json ADDED
@@ -0,0 +1,1202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.53044581,
3
+ "best_model_checkpoint": "/fs/clip-projects/geoguesser/vlms/cpm/output/minicpm-v-v2_6-chat/v3-20241108-065955/checkpoint-534",
4
+ "epoch": 1.9962616822429906,
5
+ "eval_steps": 50,
6
+ "global_step": 534,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "acc": 0.55045807,
13
+ "epoch": 0.003738317757009346,
14
+ "grad_norm": 1.613571286201477,
15
+ "learning_rate": 3.7037037037037037e-06,
16
+ "loss": 1.95084548,
17
+ "memory(GiB)": 18.97,
18
+ "step": 1,
19
+ "train_speed(iter/s)": 0.133639
20
+ },
21
+ {
22
+ "acc": 0.50916213,
23
+ "epoch": 0.018691588785046728,
24
+ "grad_norm": 1.757776141166687,
25
+ "learning_rate": 1.8518518518518518e-05,
26
+ "loss": 2.09828925,
27
+ "memory(GiB)": 19.98,
28
+ "step": 5,
29
+ "train_speed(iter/s)": 0.287053
30
+ },
31
+ {
32
+ "acc": 0.54273653,
33
+ "epoch": 0.037383177570093455,
34
+ "grad_norm": 1.2750011682510376,
35
+ "learning_rate": 3.7037037037037037e-05,
36
+ "loss": 1.98593979,
37
+ "memory(GiB)": 19.98,
38
+ "step": 10,
39
+ "train_speed(iter/s)": 0.333912
40
+ },
41
+ {
42
+ "acc": 0.53456535,
43
+ "epoch": 0.056074766355140186,
44
+ "grad_norm": 1.1665784120559692,
45
+ "learning_rate": 5.555555555555556e-05,
46
+ "loss": 1.90109081,
47
+ "memory(GiB)": 21.0,
48
+ "step": 15,
49
+ "train_speed(iter/s)": 0.352318
50
+ },
51
+ {
52
+ "acc": 0.54925652,
53
+ "epoch": 0.07476635514018691,
54
+ "grad_norm": 1.3744975328445435,
55
+ "learning_rate": 7.407407407407407e-05,
56
+ "loss": 1.76896648,
57
+ "memory(GiB)": 21.0,
58
+ "step": 20,
59
+ "train_speed(iter/s)": 0.362181
60
+ },
61
+ {
62
+ "acc": 0.5564671,
63
+ "epoch": 0.09345794392523364,
64
+ "grad_norm": 1.2068527936935425,
65
+ "learning_rate": 9.25925925925926e-05,
66
+ "loss": 1.65932693,
67
+ "memory(GiB)": 21.0,
68
+ "step": 25,
69
+ "train_speed(iter/s)": 0.368232
70
+ },
71
+ {
72
+ "acc": 0.55754151,
73
+ "epoch": 0.11214953271028037,
74
+ "grad_norm": 0.974327802658081,
75
+ "learning_rate": 9.999136119166803e-05,
76
+ "loss": 1.7156683,
77
+ "memory(GiB)": 22.04,
78
+ "step": 30,
79
+ "train_speed(iter/s)": 0.3718
80
+ },
81
+ {
82
+ "acc": 0.554459,
83
+ "epoch": 0.1308411214953271,
84
+ "grad_norm": 1.1023714542388916,
85
+ "learning_rate": 9.99385792841537e-05,
86
+ "loss": 1.76657219,
87
+ "memory(GiB)": 22.04,
88
+ "step": 35,
89
+ "train_speed(iter/s)": 0.37456
90
+ },
91
+ {
92
+ "acc": 0.55972748,
93
+ "epoch": 0.14953271028037382,
94
+ "grad_norm": 1.0901970863342285,
95
+ "learning_rate": 9.983786540671051e-05,
96
+ "loss": 1.64107857,
97
+ "memory(GiB)": 23.08,
98
+ "step": 40,
99
+ "train_speed(iter/s)": 0.376463
100
+ },
101
+ {
102
+ "acc": 0.5773067,
103
+ "epoch": 0.16822429906542055,
104
+ "grad_norm": 1.043094277381897,
105
+ "learning_rate": 9.968931622637652e-05,
106
+ "loss": 1.62755222,
107
+ "memory(GiB)": 23.08,
108
+ "step": 45,
109
+ "train_speed(iter/s)": 0.378202
110
+ },
111
+ {
112
+ "acc": 0.5643084,
113
+ "epoch": 0.18691588785046728,
114
+ "grad_norm": 1.1167865991592407,
115
+ "learning_rate": 9.949307432339625e-05,
116
+ "loss": 1.67872505,
117
+ "memory(GiB)": 23.08,
118
+ "step": 50,
119
+ "train_speed(iter/s)": 0.379293
120
+ },
121
+ {
122
+ "epoch": 0.18691588785046728,
123
+ "eval_acc": 0.5715999400209927,
124
+ "eval_loss": 1.6141570806503296,
125
+ "eval_runtime": 19.3426,
126
+ "eval_samples_per_second": 2.585,
127
+ "eval_steps_per_second": 2.585,
128
+ "step": 50
129
+ },
130
+ {
131
+ "acc": 0.57612801,
132
+ "epoch": 0.205607476635514,
133
+ "grad_norm": 1.0558923482894897,
134
+ "learning_rate": 9.924932805436949e-05,
135
+ "loss": 1.57275066,
136
+ "memory(GiB)": 23.08,
137
+ "step": 55,
138
+ "train_speed(iter/s)": 0.333772
139
+ },
140
+ {
141
+ "acc": 0.56456318,
142
+ "epoch": 0.22429906542056074,
143
+ "grad_norm": 1.2267849445343018,
144
+ "learning_rate": 9.895831137146318e-05,
145
+ "loss": 1.62593994,
146
+ "memory(GiB)": 23.08,
147
+ "step": 60,
148
+ "train_speed(iter/s)": 0.337909
149
+ },
150
+ {
151
+ "acc": 0.57293587,
152
+ "epoch": 0.24299065420560748,
153
+ "grad_norm": 1.203539252281189,
154
+ "learning_rate": 9.862030359785981e-05,
155
+ "loss": 1.62202415,
156
+ "memory(GiB)": 23.08,
157
+ "step": 65,
158
+ "train_speed(iter/s)": 0.341402
159
+ },
160
+ {
161
+ "acc": 0.56972237,
162
+ "epoch": 0.2616822429906542,
163
+ "grad_norm": 1.1968094110488892,
164
+ "learning_rate": 9.82356291596578e-05,
165
+ "loss": 1.62049065,
166
+ "memory(GiB)": 23.08,
167
+ "step": 70,
168
+ "train_speed(iter/s)": 0.344462
169
+ },
170
+ {
171
+ "acc": 0.56793709,
172
+ "epoch": 0.2803738317757009,
173
+ "grad_norm": 1.1416860818862915,
174
+ "learning_rate": 9.780465727448149e-05,
175
+ "loss": 1.68797512,
176
+ "memory(GiB)": 24.13,
177
+ "step": 75,
178
+ "train_speed(iter/s)": 0.346849
179
+ },
180
+ {
181
+ "acc": 0.58297682,
182
+ "epoch": 0.29906542056074764,
183
+ "grad_norm": 1.1054881811141968,
184
+ "learning_rate": 9.732780159709912e-05,
185
+ "loss": 1.57775593,
186
+ "memory(GiB)": 24.13,
187
+ "step": 80,
188
+ "train_speed(iter/s)": 0.349274
189
+ },
190
+ {
191
+ "acc": 0.57728238,
192
+ "epoch": 0.3177570093457944,
193
+ "grad_norm": 1.1610243320465088,
194
+ "learning_rate": 9.680551982238942e-05,
195
+ "loss": 1.59094667,
196
+ "memory(GiB)": 24.13,
197
+ "step": 85,
198
+ "train_speed(iter/s)": 0.351479
199
+ },
200
+ {
201
+ "acc": 0.5719296,
202
+ "epoch": 0.3364485981308411,
203
+ "grad_norm": 1.1896332502365112,
204
+ "learning_rate": 9.623831324603754e-05,
205
+ "loss": 1.65499535,
206
+ "memory(GiB)": 24.13,
207
+ "step": 90,
208
+ "train_speed(iter/s)": 0.353415
209
+ },
210
+ {
211
+ "acc": 0.58935571,
212
+ "epoch": 0.35514018691588783,
213
+ "grad_norm": 1.0724711418151855,
214
+ "learning_rate": 9.562672628338233e-05,
215
+ "loss": 1.58491564,
216
+ "memory(GiB)": 24.13,
217
+ "step": 95,
218
+ "train_speed(iter/s)": 0.355195
219
+ },
220
+ {
221
+ "acc": 0.56192031,
222
+ "epoch": 0.37383177570093457,
223
+ "grad_norm": 1.0967603921890259,
224
+ "learning_rate": 9.497134594687634e-05,
225
+ "loss": 1.68490829,
226
+ "memory(GiB)": 24.13,
227
+ "step": 100,
228
+ "train_speed(iter/s)": 0.356737
229
+ },
230
+ {
231
+ "epoch": 0.37383177570093457,
232
+ "eval_acc": 0.5846453741190584,
233
+ "eval_loss": 1.5773682594299316,
234
+ "eval_runtime": 19.3689,
235
+ "eval_samples_per_second": 2.581,
236
+ "eval_steps_per_second": 2.581,
237
+ "step": 100
238
+ },
239
+ {
240
+ "acc": 0.58094668,
241
+ "epoch": 0.3925233644859813,
242
+ "grad_norm": 1.0203641653060913,
243
+ "learning_rate": 9.42728012826605e-05,
244
+ "loss": 1.63088989,
245
+ "memory(GiB)": 24.13,
246
+ "step": 105,
247
+ "train_speed(iter/s)": 0.334857
248
+ },
249
+ {
250
+ "acc": 0.56821561,
251
+ "epoch": 0.411214953271028,
252
+ "grad_norm": 1.0685813426971436,
253
+ "learning_rate": 9.353176276679396e-05,
254
+ "loss": 1.67582684,
255
+ "memory(GiB)": 24.13,
256
+ "step": 110,
257
+ "train_speed(iter/s)": 0.33707
258
+ },
259
+ {
260
+ "acc": 0.56454325,
261
+ "epoch": 0.42990654205607476,
262
+ "grad_norm": 1.1684739589691162,
263
+ "learning_rate": 9.274894166171888e-05,
264
+ "loss": 1.63463154,
265
+ "memory(GiB)": 24.13,
266
+ "step": 115,
267
+ "train_speed(iter/s)": 0.33915
268
+ },
269
+ {
270
+ "acc": 0.57855415,
271
+ "epoch": 0.4485981308411215,
272
+ "grad_norm": 1.077269434928894,
273
+ "learning_rate": 9.192508933357753e-05,
274
+ "loss": 1.67274055,
275
+ "memory(GiB)": 24.13,
276
+ "step": 120,
277
+ "train_speed(iter/s)": 0.341042
278
+ },
279
+ {
280
+ "acc": 0.57130666,
281
+ "epoch": 0.4672897196261682,
282
+ "grad_norm": 1.066927194595337,
283
+ "learning_rate": 9.106099653103728e-05,
284
+ "loss": 1.57755241,
285
+ "memory(GiB)": 24.13,
286
+ "step": 125,
287
+ "train_speed(iter/s)": 0.34276
288
+ },
289
+ {
290
+ "acc": 0.57805595,
291
+ "epoch": 0.48598130841121495,
292
+ "grad_norm": 1.214378833770752,
293
+ "learning_rate": 9.015749262631536e-05,
294
+ "loss": 1.5515789,
295
+ "memory(GiB)": 24.13,
296
+ "step": 130,
297
+ "train_speed(iter/s)": 0.344333
298
+ },
299
+ {
300
+ "acc": 0.58489175,
301
+ "epoch": 0.5046728971962616,
302
+ "grad_norm": 1.2008962631225586,
303
+ "learning_rate": 8.921544481913218e-05,
304
+ "loss": 1.60749855,
305
+ "memory(GiB)": 24.13,
306
+ "step": 135,
307
+ "train_speed(iter/s)": 0.345892
308
+ },
309
+ {
310
+ "acc": 0.55853381,
311
+ "epoch": 0.5233644859813084,
312
+ "grad_norm": 1.1639275550842285,
313
+ "learning_rate": 8.823575730435693e-05,
314
+ "loss": 1.63589153,
315
+ "memory(GiB)": 25.22,
316
+ "step": 140,
317
+ "train_speed(iter/s)": 0.3473
318
+ },
319
+ {
320
+ "acc": 0.57799473,
321
+ "epoch": 0.5420560747663551,
322
+ "grad_norm": 1.0502570867538452,
323
+ "learning_rate": 8.721937040414481e-05,
324
+ "loss": 1.55451593,
325
+ "memory(GiB)": 25.22,
326
+ "step": 145,
327
+ "train_speed(iter/s)": 0.348596
328
+ },
329
+ {
330
+ "acc": 0.56694794,
331
+ "epoch": 0.5607476635514018,
332
+ "grad_norm": 1.2373900413513184,
333
+ "learning_rate": 8.616725966539832e-05,
334
+ "loss": 1.64150391,
335
+ "memory(GiB)": 25.22,
336
+ "step": 150,
337
+ "train_speed(iter/s)": 0.349855
338
+ },
339
+ {
340
+ "epoch": 0.5607476635514018,
341
+ "eval_acc": 0.5879442195231669,
342
+ "eval_loss": 1.5577113628387451,
343
+ "eval_runtime": 19.4081,
344
+ "eval_samples_per_second": 2.576,
345
+ "eval_steps_per_second": 2.576,
346
+ "step": 150
347
+ },
348
+ {
349
+ "acc": 0.57931132,
350
+ "epoch": 0.5794392523364486,
351
+ "grad_norm": 1.148728847503662,
352
+ "learning_rate": 8.508043492341944e-05,
353
+ "loss": 1.59622688,
354
+ "memory(GiB)": 25.22,
355
+ "step": 155,
356
+ "train_speed(iter/s)": 0.335587
357
+ },
358
+ {
359
+ "acc": 0.57598162,
360
+ "epoch": 0.5981308411214953,
361
+ "grad_norm": 1.1714155673980713,
362
+ "learning_rate": 8.395993933265101e-05,
363
+ "loss": 1.63730679,
364
+ "memory(GiB)": 25.22,
365
+ "step": 160,
366
+ "train_speed(iter/s)": 0.337039
367
+ },
368
+ {
369
+ "acc": 0.56635065,
370
+ "epoch": 0.616822429906542,
371
+ "grad_norm": 1.000349521636963,
372
+ "learning_rate": 8.280684836543794e-05,
373
+ "loss": 1.61145477,
374
+ "memory(GiB)": 25.22,
375
+ "step": 165,
376
+ "train_speed(iter/s)": 0.338357
377
+ },
378
+ {
379
+ "acc": 0.57298255,
380
+ "epoch": 0.6355140186915887,
381
+ "grad_norm": 1.1170839071273804,
382
+ "learning_rate": 8.162226877976887e-05,
383
+ "loss": 1.59238987,
384
+ "memory(GiB)": 25.22,
385
+ "step": 170,
386
+ "train_speed(iter/s)": 0.339705
387
+ },
388
+ {
389
+ "acc": 0.58343244,
390
+ "epoch": 0.6542056074766355,
391
+ "grad_norm": 1.0932862758636475,
392
+ "learning_rate": 8.040733755698955e-05,
393
+ "loss": 1.58805714,
394
+ "memory(GiB)": 25.22,
395
+ "step": 175,
396
+ "train_speed(iter/s)": 0.340942
397
+ },
398
+ {
399
+ "acc": 0.5746861,
400
+ "epoch": 0.6728971962616822,
401
+ "grad_norm": 1.147817850112915,
402
+ "learning_rate": 7.916322081050709e-05,
403
+ "loss": 1.58162947,
404
+ "memory(GiB)": 25.22,
405
+ "step": 180,
406
+ "train_speed(iter/s)": 0.342139
407
+ },
408
+ {
409
+ "acc": 0.57306166,
410
+ "epoch": 0.6915887850467289,
411
+ "grad_norm": 1.076221227645874,
412
+ "learning_rate": 7.789111266653285e-05,
413
+ "loss": 1.58194542,
414
+ "memory(GiB)": 25.22,
415
+ "step": 185,
416
+ "train_speed(iter/s)": 0.343249
417
+ },
418
+ {
419
+ "acc": 0.58896809,
420
+ "epoch": 0.7102803738317757,
421
+ "grad_norm": 1.1743425130844116,
422
+ "learning_rate": 7.659223411793798e-05,
423
+ "loss": 1.53554783,
424
+ "memory(GiB)": 25.22,
425
+ "step": 190,
426
+ "train_speed(iter/s)": 0.344334
427
+ },
428
+ {
429
+ "acc": 0.57240195,
430
+ "epoch": 0.7289719626168224,
431
+ "grad_norm": 1.0945876836776733,
432
+ "learning_rate": 7.526783185232207e-05,
433
+ "loss": 1.59289436,
434
+ "memory(GiB)": 25.22,
435
+ "step": 195,
436
+ "train_speed(iter/s)": 0.345336
437
+ },
438
+ {
439
+ "acc": 0.5794302,
440
+ "epoch": 0.7476635514018691,
441
+ "grad_norm": 1.1055279970169067,
442
+ "learning_rate": 7.391917705541927e-05,
443
+ "loss": 1.621562,
444
+ "memory(GiB)": 25.22,
445
+ "step": 200,
446
+ "train_speed(iter/s)": 0.346324
447
+ },
448
+ {
449
+ "epoch": 0.7476635514018691,
450
+ "eval_acc": 0.5871944819313241,
451
+ "eval_loss": 1.5511505603790283,
452
+ "eval_runtime": 19.403,
453
+ "eval_samples_per_second": 2.577,
454
+ "eval_steps_per_second": 2.577,
455
+ "step": 200
456
+ },
457
+ {
458
+ "acc": 0.56806307,
459
+ "epoch": 0.7663551401869159,
460
+ "grad_norm": 1.017061471939087,
461
+ "learning_rate": 7.254756419099074e-05,
462
+ "loss": 1.6349041,
463
+ "memory(GiB)": 25.22,
464
+ "step": 205,
465
+ "train_speed(iter/s)": 0.335703
466
+ },
467
+ {
468
+ "acc": 0.56848702,
469
+ "epoch": 0.7850467289719626,
470
+ "grad_norm": 1.1138246059417725,
471
+ "learning_rate": 7.115430975837457e-05,
472
+ "loss": 1.60775127,
473
+ "memory(GiB)": 25.22,
474
+ "step": 210,
475
+ "train_speed(iter/s)": 0.336869
476
+ },
477
+ {
478
+ "acc": 0.58039517,
479
+ "epoch": 0.8037383177570093,
480
+ "grad_norm": 1.2759917974472046,
481
+ "learning_rate": 6.974075102888536e-05,
482
+ "loss": 1.59430618,
483
+ "memory(GiB)": 25.22,
484
+ "step": 215,
485
+ "train_speed(iter/s)": 0.337979
486
+ },
487
+ {
488
+ "acc": 0.56688986,
489
+ "epoch": 0.822429906542056,
490
+ "grad_norm": 1.259171724319458,
491
+ "learning_rate": 6.830824476227646e-05,
492
+ "loss": 1.61376076,
493
+ "memory(GiB)": 25.22,
494
+ "step": 220,
495
+ "train_speed(iter/s)": 0.339033
496
+ },
497
+ {
498
+ "acc": 0.58966751,
499
+ "epoch": 0.8411214953271028,
500
+ "grad_norm": 1.1861165761947632,
501
+ "learning_rate": 6.685816590449708e-05,
502
+ "loss": 1.56043501,
503
+ "memory(GiB)": 25.22,
504
+ "step": 225,
505
+ "train_speed(iter/s)": 0.340035
506
+ },
507
+ {
508
+ "acc": 0.58497729,
509
+ "epoch": 0.8598130841121495,
510
+ "grad_norm": 1.1015815734863281,
511
+ "learning_rate": 6.539190626799366e-05,
512
+ "loss": 1.57877932,
513
+ "memory(GiB)": 25.22,
514
+ "step": 230,
515
+ "train_speed(iter/s)": 0.341016
516
+ },
517
+ {
518
+ "acc": 0.58781128,
519
+ "epoch": 0.8785046728971962,
520
+ "grad_norm": 1.2020829916000366,
521
+ "learning_rate": 6.391087319582264e-05,
522
+ "loss": 1.57381382,
523
+ "memory(GiB)": 25.22,
524
+ "step": 235,
525
+ "train_speed(iter/s)": 0.341901
526
+ },
527
+ {
528
+ "acc": 0.57930512,
529
+ "epoch": 0.897196261682243,
530
+ "grad_norm": 1.1107765436172485,
531
+ "learning_rate": 6.241648821085666e-05,
532
+ "loss": 1.56744556,
533
+ "memory(GiB)": 25.22,
534
+ "step": 240,
535
+ "train_speed(iter/s)": 0.338406
536
+ },
537
+ {
538
+ "acc": 0.57636843,
539
+ "epoch": 0.9158878504672897,
540
+ "grad_norm": 1.1953294277191162,
541
+ "learning_rate": 6.0910185651380626e-05,
542
+ "loss": 1.53063288,
543
+ "memory(GiB)": 25.22,
544
+ "step": 245,
545
+ "train_speed(iter/s)": 0.339394
546
+ },
547
+ {
548
+ "acc": 0.57735896,
549
+ "epoch": 0.9345794392523364,
550
+ "grad_norm": 1.1470882892608643,
551
+ "learning_rate": 5.939341129438739e-05,
552
+ "loss": 1.64163361,
553
+ "memory(GiB)": 25.22,
554
+ "step": 250,
555
+ "train_speed(iter/s)": 0.340265
556
+ },
557
+ {
558
+ "epoch": 0.9345794392523364,
559
+ "eval_acc": 0.5865946918578497,
560
+ "eval_loss": 1.539453148841858,
561
+ "eval_runtime": 19.2796,
562
+ "eval_samples_per_second": 2.593,
563
+ "eval_steps_per_second": 2.593,
564
+ "step": 250
565
+ },
566
+ {
567
+ "acc": 0.55690722,
568
+ "epoch": 0.9532710280373832,
569
+ "grad_norm": 1.225205421447754,
570
+ "learning_rate": 5.786762096789431e-05,
571
+ "loss": 1.68273106,
572
+ "memory(GiB)": 25.22,
573
+ "step": 255,
574
+ "train_speed(iter/s)": 0.332148
575
+ },
576
+ {
577
+ "acc": 0.56964278,
578
+ "epoch": 0.9719626168224299,
579
+ "grad_norm": 1.0283957719802856,
580
+ "learning_rate": 5.633427915361261e-05,
581
+ "loss": 1.64168797,
582
+ "memory(GiB)": 25.22,
583
+ "step": 260,
584
+ "train_speed(iter/s)": 0.333054
585
+ },
586
+ {
587
+ "acc": 0.55817146,
588
+ "epoch": 0.9906542056074766,
589
+ "grad_norm": 1.0989590883255005,
590
+ "learning_rate": 5.479485758131089e-05,
591
+ "loss": 1.61564522,
592
+ "memory(GiB)": 25.22,
593
+ "step": 265,
594
+ "train_speed(iter/s)": 0.333997
595
+ },
596
+ {
597
+ "acc": 0.59431157,
598
+ "epoch": 1.0093457943925233,
599
+ "grad_norm": 1.0156042575836182,
600
+ "learning_rate": 5.325083381622165e-05,
601
+ "loss": 1.54522419,
602
+ "memory(GiB)": 25.22,
603
+ "step": 270,
604
+ "train_speed(iter/s)": 0.334781
605
+ },
606
+ {
607
+ "acc": 0.59816217,
608
+ "epoch": 1.02803738317757,
609
+ "grad_norm": 1.1333341598510742,
610
+ "learning_rate": 5.1703689840846945e-05,
611
+ "loss": 1.43400564,
612
+ "memory(GiB)": 25.22,
613
+ "step": 275,
614
+ "train_speed(iter/s)": 0.335687
615
+ },
616
+ {
617
+ "acc": 0.5830586,
618
+ "epoch": 1.0467289719626167,
619
+ "grad_norm": 1.289959192276001,
620
+ "learning_rate": 5.01549106325243e-05,
621
+ "loss": 1.52126703,
622
+ "memory(GiB)": 25.22,
623
+ "step": 280,
624
+ "train_speed(iter/s)": 0.336526
625
+ },
626
+ {
627
+ "acc": 0.62401681,
628
+ "epoch": 1.0654205607476634,
629
+ "grad_norm": 1.3222240209579468,
630
+ "learning_rate": 4.860598273811792e-05,
631
+ "loss": 1.36042976,
632
+ "memory(GiB)": 25.22,
633
+ "step": 285,
634
+ "train_speed(iter/s)": 0.337368
635
+ },
636
+ {
637
+ "acc": 0.57799129,
638
+ "epoch": 1.0841121495327102,
639
+ "grad_norm": 1.261518120765686,
640
+ "learning_rate": 4.705839284720376e-05,
641
+ "loss": 1.48789501,
642
+ "memory(GiB)": 25.22,
643
+ "step": 290,
644
+ "train_speed(iter/s)": 0.338162
645
+ },
646
+ {
647
+ "acc": 0.60333071,
648
+ "epoch": 1.102803738317757,
649
+ "grad_norm": 1.351491928100586,
650
+ "learning_rate": 4.55136263651172e-05,
651
+ "loss": 1.49587126,
652
+ "memory(GiB)": 25.22,
653
+ "step": 295,
654
+ "train_speed(iter/s)": 0.338912
655
+ },
656
+ {
657
+ "acc": 0.61315393,
658
+ "epoch": 1.1214953271028036,
659
+ "grad_norm": 1.6314853429794312,
660
+ "learning_rate": 4.397316598723385e-05,
661
+ "loss": 1.43088112,
662
+ "memory(GiB)": 25.22,
663
+ "step": 300,
664
+ "train_speed(iter/s)": 0.339653
665
+ },
666
+ {
667
+ "epoch": 1.1214953271028036,
668
+ "eval_acc": 0.5912430649272754,
669
+ "eval_loss": 1.5384690761566162,
670
+ "eval_runtime": 19.3435,
671
+ "eval_samples_per_second": 2.585,
672
+ "eval_steps_per_second": 2.585,
673
+ "step": 300
674
+ },
675
+ {
676
+ "acc": 0.60021753,
677
+ "epoch": 1.1401869158878504,
678
+ "grad_norm": 1.4269309043884277,
679
+ "learning_rate": 4.243849027585096e-05,
680
+ "loss": 1.47466078,
681
+ "memory(GiB)": 25.22,
682
+ "step": 305,
683
+ "train_speed(iter/s)": 0.332888
684
+ },
685
+ {
686
+ "acc": 0.58526664,
687
+ "epoch": 1.158878504672897,
688
+ "grad_norm": 1.4827656745910645,
689
+ "learning_rate": 4.0911072241036194e-05,
690
+ "loss": 1.53870859,
691
+ "memory(GiB)": 25.22,
692
+ "step": 310,
693
+ "train_speed(iter/s)": 0.333651
694
+ },
695
+ {
696
+ "acc": 0.6124382,
697
+ "epoch": 1.1775700934579438,
698
+ "grad_norm": 1.428358793258667,
699
+ "learning_rate": 3.9392377926805226e-05,
700
+ "loss": 1.44478369,
701
+ "memory(GiB)": 25.22,
702
+ "step": 315,
703
+ "train_speed(iter/s)": 0.334437
704
+ },
705
+ {
706
+ "acc": 0.58961325,
707
+ "epoch": 1.1962616822429906,
708
+ "grad_norm": 1.4965459108352661,
709
+ "learning_rate": 3.788386500398583e-05,
710
+ "loss": 1.49487057,
711
+ "memory(GiB)": 25.22,
712
+ "step": 320,
713
+ "train_speed(iter/s)": 0.335206
714
+ },
715
+ {
716
+ "acc": 0.58942304,
717
+ "epoch": 1.2149532710280373,
718
+ "grad_norm": 1.3801062107086182,
719
+ "learning_rate": 3.6386981371118355e-05,
720
+ "loss": 1.44996719,
721
+ "memory(GiB)": 25.22,
722
+ "step": 325,
723
+ "train_speed(iter/s)": 0.335913
724
+ },
725
+ {
726
+ "acc": 0.5930171,
727
+ "epoch": 1.233644859813084,
728
+ "grad_norm": 1.550034523010254,
729
+ "learning_rate": 3.49031637647361e-05,
730
+ "loss": 1.4618269,
731
+ "memory(GiB)": 25.22,
732
+ "step": 330,
733
+ "train_speed(iter/s)": 0.336661
734
+ },
735
+ {
736
+ "acc": 0.61733942,
737
+ "epoch": 1.2523364485981308,
738
+ "grad_norm": 1.467942237854004,
739
+ "learning_rate": 3.343383638035902e-05,
740
+ "loss": 1.37453032,
741
+ "memory(GiB)": 25.22,
742
+ "step": 335,
743
+ "train_speed(iter/s)": 0.337353
744
+ },
745
+ {
746
+ "acc": 0.60188942,
747
+ "epoch": 1.2710280373831775,
748
+ "grad_norm": 1.571946620941162,
749
+ "learning_rate": 3.1980409505524544e-05,
750
+ "loss": 1.42003136,
751
+ "memory(GiB)": 25.22,
752
+ "step": 340,
753
+ "train_speed(iter/s)": 0.33804
754
+ },
755
+ {
756
+ "acc": 0.60350924,
757
+ "epoch": 1.2897196261682242,
758
+ "grad_norm": 1.773979663848877,
759
+ "learning_rate": 3.054427816616773e-05,
760
+ "loss": 1.40252638,
761
+ "memory(GiB)": 25.22,
762
+ "step": 345,
763
+ "train_speed(iter/s)": 0.338706
764
+ },
765
+ {
766
+ "acc": 0.59472365,
767
+ "epoch": 1.308411214953271,
768
+ "grad_norm": 1.5724050998687744,
769
+ "learning_rate": 2.91268207876494e-05,
770
+ "loss": 1.46759853,
771
+ "memory(GiB)": 25.22,
772
+ "step": 350,
773
+ "train_speed(iter/s)": 0.339365
774
+ },
775
+ {
776
+ "epoch": 1.308411214953271,
777
+ "eval_acc": 0.5880941670415355,
778
+ "eval_loss": 1.5372613668441772,
779
+ "eval_runtime": 19.309,
780
+ "eval_samples_per_second": 2.589,
781
+ "eval_steps_per_second": 2.589,
782
+ "step": 350
783
+ },
784
+ {
785
+ "acc": 0.60742517,
786
+ "epoch": 1.3271028037383177,
787
+ "grad_norm": 1.6696964502334595,
788
+ "learning_rate": 2.7729397871718304e-05,
789
+ "loss": 1.41251793,
790
+ "memory(GiB)": 25.22,
791
+ "step": 355,
792
+ "train_speed(iter/s)": 0.333524
793
+ },
794
+ {
795
+ "acc": 0.59874382,
796
+ "epoch": 1.3457943925233644,
797
+ "grad_norm": 1.631954550743103,
798
+ "learning_rate": 2.635335069067617e-05,
799
+ "loss": 1.43565807,
800
+ "memory(GiB)": 25.22,
801
+ "step": 360,
802
+ "train_speed(iter/s)": 0.33422
803
+ },
804
+ {
805
+ "acc": 0.62206426,
806
+ "epoch": 1.3644859813084111,
807
+ "grad_norm": 1.6964654922485352,
808
+ "learning_rate": 2.500000000000001e-05,
809
+ "loss": 1.40124302,
810
+ "memory(GiB)": 25.22,
811
+ "step": 365,
812
+ "train_speed(iter/s)": 0.334875
813
+ },
814
+ {
815
+ "acc": 0.60629749,
816
+ "epoch": 1.3831775700934579,
817
+ "grad_norm": 1.4727040529251099,
818
+ "learning_rate": 2.367064477065652e-05,
819
+ "loss": 1.4464427,
820
+ "memory(GiB)": 25.22,
821
+ "step": 370,
822
+ "train_speed(iter/s)": 0.335544
823
+ },
824
+ {
825
+ "acc": 0.60391579,
826
+ "epoch": 1.4018691588785046,
827
+ "grad_norm": 1.7280242443084717,
828
+ "learning_rate": 2.2366560942325832e-05,
829
+ "loss": 1.41403561,
830
+ "memory(GiB)": 25.22,
831
+ "step": 375,
832
+ "train_speed(iter/s)": 0.336204
833
+ },
834
+ {
835
+ "acc": 0.60924401,
836
+ "epoch": 1.4205607476635513,
837
+ "grad_norm": 1.7434614896774292,
838
+ "learning_rate": 2.108900019873103e-05,
839
+ "loss": 1.44466,
840
+ "memory(GiB)": 25.22,
841
+ "step": 380,
842
+ "train_speed(iter/s)": 0.336821
843
+ },
844
+ {
845
+ "acc": 0.62293906,
846
+ "epoch": 1.439252336448598,
847
+ "grad_norm": 1.5931147336959839,
848
+ "learning_rate": 1.983918876624902e-05,
849
+ "loss": 1.36733408,
850
+ "memory(GiB)": 25.22,
851
+ "step": 385,
852
+ "train_speed(iter/s)": 0.337406
853
+ },
854
+ {
855
+ "acc": 0.60391083,
856
+ "epoch": 1.4579439252336448,
857
+ "grad_norm": 1.8774250745773315,
858
+ "learning_rate": 1.8618326236955907e-05,
859
+ "loss": 1.47434216,
860
+ "memory(GiB)": 25.22,
861
+ "step": 390,
862
+ "train_speed(iter/s)": 0.337991
863
+ },
864
+ {
865
+ "acc": 0.59494829,
866
+ "epoch": 1.4766355140186915,
867
+ "grad_norm": 1.6903536319732666,
868
+ "learning_rate": 1.7427584417236194e-05,
869
+ "loss": 1.4961113,
870
+ "memory(GiB)": 25.22,
871
+ "step": 395,
872
+ "train_speed(iter/s)": 0.338556
873
+ },
874
+ {
875
+ "acc": 0.61160607,
876
+ "epoch": 1.4953271028037383,
877
+ "grad_norm": 1.6840981245040894,
878
+ "learning_rate": 1.626810620306163e-05,
879
+ "loss": 1.3965476,
880
+ "memory(GiB)": 25.22,
881
+ "step": 400,
882
+ "train_speed(iter/s)": 0.339121
883
+ },
884
+ {
885
+ "epoch": 1.4953271028037383,
886
+ "eval_acc": 0.5915429599640126,
887
+ "eval_loss": 1.5317269563674927,
888
+ "eval_runtime": 19.3286,
889
+ "eval_samples_per_second": 2.587,
890
+ "eval_steps_per_second": 2.587,
891
+ "step": 400
892
+ },
893
+ {
894
+ "acc": 0.59379582,
895
+ "epoch": 1.514018691588785,
896
+ "grad_norm": 1.6205955743789673,
897
+ "learning_rate": 1.5141004483018323e-05,
898
+ "loss": 1.44787579,
899
+ "memory(GiB)": 25.22,
900
+ "step": 405,
901
+ "train_speed(iter/s)": 0.333982
902
+ },
903
+ {
904
+ "acc": 0.61180854,
905
+ "epoch": 1.5327102803738317,
906
+ "grad_norm": 1.9331731796264648,
907
+ "learning_rate": 1.4047361070135995e-05,
908
+ "loss": 1.46465635,
909
+ "memory(GiB)": 25.22,
910
+ "step": 410,
911
+ "train_speed(iter/s)": 0.334588
912
+ },
913
+ {
914
+ "acc": 0.59493322,
915
+ "epoch": 1.5514018691588785,
916
+ "grad_norm": 1.579399824142456,
917
+ "learning_rate": 1.2988225663543602e-05,
918
+ "loss": 1.51334658,
919
+ "memory(GiB)": 25.22,
920
+ "step": 415,
921
+ "train_speed(iter/s)": 0.335141
922
+ },
923
+ {
924
+ "acc": 0.60085406,
925
+ "epoch": 1.5700934579439252,
926
+ "grad_norm": 1.7813141345977783,
927
+ "learning_rate": 1.1964614840949002e-05,
928
+ "loss": 1.44739676,
929
+ "memory(GiB)": 25.22,
930
+ "step": 420,
931
+ "train_speed(iter/s)": 0.335707
932
+ },
933
+ {
934
+ "acc": 0.59427462,
935
+ "epoch": 1.588785046728972,
936
+ "grad_norm": 1.6966168880462646,
937
+ "learning_rate": 1.097751108290867e-05,
938
+ "loss": 1.47780085,
939
+ "memory(GiB)": 25.22,
940
+ "step": 425,
941
+ "train_speed(iter/s)": 0.336254
942
+ },
943
+ {
944
+ "acc": 0.623209,
945
+ "epoch": 1.6074766355140186,
946
+ "grad_norm": 1.801193118095398,
947
+ "learning_rate": 1.0027861829824952e-05,
948
+ "loss": 1.37850494,
949
+ "memory(GiB)": 25.22,
950
+ "step": 430,
951
+ "train_speed(iter/s)": 0.336818
952
+ },
953
+ {
954
+ "acc": 0.60559597,
955
+ "epoch": 1.6261682242990654,
956
+ "grad_norm": 1.694907307624817,
957
+ "learning_rate": 9.11657857257509e-06,
958
+ "loss": 1.47119045,
959
+ "memory(GiB)": 25.22,
960
+ "step": 435,
961
+ "train_speed(iter/s)": 0.337357
962
+ },
963
+ {
964
+ "acc": 0.60444808,
965
+ "epoch": 1.644859813084112,
966
+ "grad_norm": 1.8144758939743042,
967
+ "learning_rate": 8.244535977645585e-06,
968
+ "loss": 1.46991854,
969
+ "memory(GiB)": 25.22,
970
+ "step": 440,
971
+ "train_speed(iter/s)": 0.337883
972
+ },
973
+ {
974
+ "acc": 0.61087198,
975
+ "epoch": 1.6635514018691588,
976
+ "grad_norm": 1.8258271217346191,
977
+ "learning_rate": 7.412571047611155e-06,
978
+ "loss": 1.39234638,
979
+ "memory(GiB)": 25.22,
980
+ "step": 445,
981
+ "train_speed(iter/s)": 0.3384
982
+ },
983
+ {
984
+ "acc": 0.60143399,
985
+ "epoch": 1.6822429906542056,
986
+ "grad_norm": 1.8941428661346436,
987
+ "learning_rate": 6.621482317764105e-06,
988
+ "loss": 1.43530188,
989
+ "memory(GiB)": 25.22,
990
+ "step": 450,
991
+ "train_speed(iter/s)": 0.338913
992
+ },
993
+ {
994
+ "epoch": 1.6822429906542056,
995
+ "eval_acc": 0.5921427500374868,
996
+ "eval_loss": 1.532676100730896,
997
+ "eval_runtime": 19.3483,
998
+ "eval_samples_per_second": 2.584,
999
+ "eval_steps_per_second": 2.584,
1000
+ "step": 450
1001
+ },
1002
+ {
1003
+ "acc": 0.61480565,
1004
+ "epoch": 1.7009345794392523,
1005
+ "grad_norm": 1.939122200012207,
1006
+ "learning_rate": 5.872029089665587e-06,
1007
+ "loss": 1.39058199,
1008
+ "memory(GiB)": 25.22,
1009
+ "step": 455,
1010
+ "train_speed(iter/s)": 0.33434
1011
+ },
1012
+ {
1013
+ "acc": 0.60005183,
1014
+ "epoch": 1.719626168224299,
1015
+ "grad_norm": 1.91712486743927,
1016
+ "learning_rate": 5.164930702353782e-06,
1017
+ "loss": 1.45630856,
1018
+ "memory(GiB)": 25.22,
1019
+ "step": 460,
1020
+ "train_speed(iter/s)": 0.334872
1021
+ },
1022
+ {
1023
+ "acc": 0.59249868,
1024
+ "epoch": 1.7383177570093458,
1025
+ "grad_norm": 1.6661227941513062,
1026
+ "learning_rate": 4.500865841909168e-06,
1027
+ "loss": 1.46352968,
1028
+ "memory(GiB)": 25.22,
1029
+ "step": 465,
1030
+ "train_speed(iter/s)": 0.335375
1031
+ },
1032
+ {
1033
+ "acc": 0.6019537,
1034
+ "epoch": 1.7570093457943925,
1035
+ "grad_norm": 1.7562310695648193,
1036
+ "learning_rate": 3.880471890038967e-06,
1037
+ "loss": 1.4511817,
1038
+ "memory(GiB)": 25.22,
1039
+ "step": 470,
1040
+ "train_speed(iter/s)": 0.335886
1041
+ },
1042
+ {
1043
+ "acc": 0.60008221,
1044
+ "epoch": 1.7757009345794392,
1045
+ "grad_norm": 1.6243735551834106,
1046
+ "learning_rate": 3.3043443123065286e-06,
1047
+ "loss": 1.49449492,
1048
+ "memory(GiB)": 25.22,
1049
+ "step": 475,
1050
+ "train_speed(iter/s)": 0.336362
1051
+ },
1052
+ {
1053
+ "acc": 0.59502878,
1054
+ "epoch": 1.794392523364486,
1055
+ "grad_norm": 2.1284008026123047,
1056
+ "learning_rate": 2.7730360865923956e-06,
1057
+ "loss": 1.46432257,
1058
+ "memory(GiB)": 25.22,
1059
+ "step": 480,
1060
+ "train_speed(iter/s)": 0.336856
1061
+ },
1062
+ {
1063
+ "acc": 0.58453884,
1064
+ "epoch": 1.8130841121495327,
1065
+ "grad_norm": 1.7423293590545654,
1066
+ "learning_rate": 2.287057172336021e-06,
1067
+ "loss": 1.53189554,
1068
+ "memory(GiB)": 25.22,
1069
+ "step": 485,
1070
+ "train_speed(iter/s)": 0.337332
1071
+ },
1072
+ {
1073
+ "acc": 0.59840698,
1074
+ "epoch": 1.8317757009345794,
1075
+ "grad_norm": 1.787650227546692,
1076
+ "learning_rate": 1.8468740210672076e-06,
1077
+ "loss": 1.44042816,
1078
+ "memory(GiB)": 25.22,
1079
+ "step": 490,
1080
+ "train_speed(iter/s)": 0.337808
1081
+ },
1082
+ {
1083
+ "acc": 0.60692272,
1084
+ "epoch": 1.8504672897196262,
1085
+ "grad_norm": 1.7444405555725098,
1086
+ "learning_rate": 1.4529091286973995e-06,
1087
+ "loss": 1.42584867,
1088
+ "memory(GiB)": 25.22,
1089
+ "step": 495,
1090
+ "train_speed(iter/s)": 0.338284
1091
+ },
1092
+ {
1093
+ "acc": 0.58918037,
1094
+ "epoch": 1.8691588785046729,
1095
+ "grad_norm": 1.7888891696929932,
1096
+ "learning_rate": 1.1055406300002347e-06,
1097
+ "loss": 1.48684044,
1098
+ "memory(GiB)": 25.22,
1099
+ "step": 500,
1100
+ "train_speed(iter/s)": 0.338742
1101
+ },
1102
+ {
1103
+ "epoch": 1.8691588785046729,
1104
+ "eval_acc": 0.5931923826660669,
1105
+ "eval_loss": 1.5313353538513184,
1106
+ "eval_runtime": 19.3381,
1107
+ "eval_samples_per_second": 2.586,
1108
+ "eval_steps_per_second": 2.586,
1109
+ "step": 500
1110
+ },
1111
+ {
1112
+ "acc": 0.61618199,
1113
+ "epoch": 1.8878504672897196,
1114
+ "grad_norm": 1.9864728450775146,
1115
+ "learning_rate": 8.0510193567086e-07,
1116
+ "loss": 1.43374748,
1117
+ "memory(GiB)": 25.22,
1118
+ "step": 505,
1119
+ "train_speed(iter/s)": 0.334578
1120
+ },
1121
+ {
1122
+ "acc": 0.60259299,
1123
+ "epoch": 1.9065420560747663,
1124
+ "grad_norm": 1.688388705253601,
1125
+ "learning_rate": 5.518814123121885e-07,
1126
+ "loss": 1.49144144,
1127
+ "memory(GiB)": 25.22,
1128
+ "step": 510,
1129
+ "train_speed(iter/s)": 0.335056
1130
+ },
1131
+ {
1132
+ "acc": 0.61285515,
1133
+ "epoch": 1.925233644859813,
1134
+ "grad_norm": 1.6542000770568848,
1135
+ "learning_rate": 3.4612210565528326e-07,
1136
+ "loss": 1.44376268,
1137
+ "memory(GiB)": 25.22,
1138
+ "step": 515,
1139
+ "train_speed(iter/s)": 0.335522
1140
+ },
1141
+ {
1142
+ "acc": 0.61628981,
1143
+ "epoch": 1.9439252336448598,
1144
+ "grad_norm": 2.033604145050049,
1145
+ "learning_rate": 1.8802150727962876e-07,
1146
+ "loss": 1.39236612,
1147
+ "memory(GiB)": 25.22,
1148
+ "step": 520,
1149
+ "train_speed(iter/s)": 0.335983
1150
+ },
1151
+ {
1152
+ "acc": 0.62410831,
1153
+ "epoch": 1.9626168224299065,
1154
+ "grad_norm": 1.6692521572113037,
1155
+ "learning_rate": 7.773136505700995e-08,
1156
+ "loss": 1.36968622,
1157
+ "memory(GiB)": 25.22,
1158
+ "step": 525,
1159
+ "train_speed(iter/s)": 0.336429
1160
+ },
1161
+ {
1162
+ "acc": 0.60063834,
1163
+ "epoch": 1.9813084112149533,
1164
+ "grad_norm": 1.7963093519210815,
1165
+ "learning_rate": 1.5357537501159423e-08,
1166
+ "loss": 1.46681242,
1167
+ "memory(GiB)": 25.22,
1168
+ "step": 530,
1169
+ "train_speed(iter/s)": 0.336698
1170
+ },
1171
+ {
1172
+ "epoch": 1.9962616822429906,
1173
+ "eval_acc": 0.5915429599640126,
1174
+ "eval_loss": 1.5304458141326904,
1175
+ "eval_runtime": 20.9568,
1176
+ "eval_samples_per_second": 2.386,
1177
+ "eval_steps_per_second": 2.386,
1178
+ "step": 534
1179
+ }
1180
+ ],
1181
+ "logging_steps": 5,
1182
+ "max_steps": 534,
1183
+ "num_input_tokens_seen": 0,
1184
+ "num_train_epochs": 2,
1185
+ "save_steps": 50,
1186
+ "stateful_callbacks": {
1187
+ "TrainerControl": {
1188
+ "args": {
1189
+ "should_epoch_stop": false,
1190
+ "should_evaluate": false,
1191
+ "should_log": false,
1192
+ "should_save": true,
1193
+ "should_training_stop": true
1194
+ },
1195
+ "attributes": {}
1196
+ }
1197
+ },
1198
+ "total_flos": 8.23057811304193e+16,
1199
+ "train_batch_size": 1,
1200
+ "trial_name": null,
1201
+ "trial_params": null
1202
+ }
minicpm-v-v2_6-chat/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e3af870f81f5afa455e78d0a60a99ba632dc9eca9239b594c659caed624b59bf
3
+ size 7416
qwen2-vl-7b-instruct/README.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: /fs/clip-projects/geoguesser/vlms/qwen/Qwen2-VL-7B-Instruct
3
+ library_name: peft
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+ ### Framework versions
201
+
202
+ - PEFT 0.12.0
qwen2-vl-7b-instruct/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "/fs/clip-projects/geoguesser/vlms/qwen/Qwen2-VL-7B-Instruct",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": [],
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "^(model)(?!.*(lm_head|output|emb|wte|shared)).*",
23
+ "task_type": "CAUSAL_LM",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
qwen2-vl-7b-instruct/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad3e18d84609148116c11627b4ee17df62b366c21aa774b2f935d104dc175a31
3
+ size 80792096
qwen2-vl-7b-instruct/additional_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}
qwen2-vl-7b-instruct/configuration.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "framework": "pytorch",
3
+ "task": "image-text-to-text",
4
+ "allow_remote": true,
5
+ "adapter_cfg": {
6
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/qwen/Qwen2-VL-7B-Instruct",
7
+ "model_revision": "master",
8
+ "sft_type": "lora",
9
+ "tuner_backend": "peft",
10
+ "template_type": "qwen2-vl",
11
+ "dtype": "bf16",
12
+ "system": "You are a helpful assistant."
13
+ }
14
+ }
qwen2-vl-7b-instruct/generation_config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": true,
4
+ "eos_token_id": 151645,
5
+ "max_new_tokens": 2048,
6
+ "pad_token_id": 151643,
7
+ "temperature": 0.01,
8
+ "top_k": 1,
9
+ "top_p": 0.001,
10
+ "transformers_version": "4.45.1"
11
+ }
qwen2-vl-7b-instruct/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47a76e0fedff89408fd4f5f5cc90b9d92a914f72baf530908a077c32e47e894e
3
+ size 161810282
qwen2-vl-7b-instruct/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13685db7db07a31ba75d2613df933051064ed0b3f50d2ac2fbbb543b65f88f0c
3
+ size 14244
qwen2-vl-7b-instruct/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b75bb0cab5c1ac64f1f42e99abb3e1c6d095394f6dd0c73c81aee621b476d6d9
3
+ size 1064
qwen2-vl-7b-instruct/sft_args.json ADDED
@@ -0,0 +1,247 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "qwen2-vl-7b-instruct",
3
+ "model_id_or_path": "/fs/clip-projects/geoguesser/vlms/qwen/Qwen2-VL-7B-Instruct",
4
+ "model_revision": "master",
5
+ "full_determinism": false,
6
+ "sft_type": "lora",
7
+ "freeze_parameters": [],
8
+ "freeze_vit": false,
9
+ "freeze_parameters_ratio": 0.0,
10
+ "additional_trainable_parameters": [],
11
+ "tuner_backend": "peft",
12
+ "template_type": "qwen2-vl",
13
+ "output_dir": "/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635",
14
+ "add_output_dir_suffix": true,
15
+ "ddp_backend": null,
16
+ "ddp_find_unused_parameters": null,
17
+ "ddp_broadcast_buffers": null,
18
+ "ddp_timeout": 1800,
19
+ "seed": 42,
20
+ "resume_from_checkpoint": null,
21
+ "resume_only_model": false,
22
+ "ignore_data_skip": false,
23
+ "dtype": "bf16",
24
+ "packing": false,
25
+ "train_backend": "transformers",
26
+ "tp": 1,
27
+ "pp": 1,
28
+ "min_lr": null,
29
+ "sequence_parallel": false,
30
+ "model_kwargs": null,
31
+ "loss_name": null,
32
+ "dataset": [
33
+ "train_set_a.jsonl"
34
+ ],
35
+ "val_dataset": [
36
+ "test_set_a.jsonl"
37
+ ],
38
+ "dataset_seed": 42,
39
+ "dataset_test_ratio": 0.0,
40
+ "use_loss_scale": false,
41
+ "loss_scale_config_path": "/fs/clip-projects/geoguesser/zheyuan/DPO/swift/swift/llm/agent/default_loss_scale_config.json",
42
+ "system": "You are a helpful assistant.",
43
+ "tools_prompt": "react_en",
44
+ "max_length": 2048,
45
+ "truncation_strategy": "delete",
46
+ "check_dataset_strategy": "none",
47
+ "streaming": false,
48
+ "streaming_val_size": 0,
49
+ "streaming_buffer_size": 16384,
50
+ "model_name": [
51
+ null,
52
+ null
53
+ ],
54
+ "model_author": [
55
+ null,
56
+ null
57
+ ],
58
+ "quant_method": null,
59
+ "quantization_bit": 0,
60
+ "hqq_axis": 0,
61
+ "hqq_dynamic_config_path": null,
62
+ "bnb_4bit_comp_dtype": "bf16",
63
+ "bnb_4bit_quant_type": "nf4",
64
+ "bnb_4bit_use_double_quant": true,
65
+ "bnb_4bit_quant_storage": null,
66
+ "rescale_image": -1,
67
+ "target_modules": "^(model)(?!.*(lm_head|output|emb|wte|shared)).*",
68
+ "target_regex": null,
69
+ "modules_to_save": [],
70
+ "lora_rank": 8,
71
+ "lora_alpha": 32,
72
+ "lora_dropout": 0.05,
73
+ "lora_bias_trainable": "none",
74
+ "lora_dtype": null,
75
+ "lora_lr_ratio": null,
76
+ "use_rslora": false,
77
+ "use_dora": false,
78
+ "init_lora_weights": true,
79
+ "fourier_n_frequency": 2000,
80
+ "fourier_scaling": 300.0,
81
+ "rope_scaling": null,
82
+ "boft_block_size": 4,
83
+ "boft_block_num": 0,
84
+ "boft_n_butterfly_factor": 1,
85
+ "boft_dropout": 0.0,
86
+ "vera_rank": 256,
87
+ "vera_projection_prng_key": 0,
88
+ "vera_dropout": 0.0,
89
+ "vera_d_initial": 0.1,
90
+ "adapter_act": "gelu",
91
+ "adapter_length": 128,
92
+ "use_galore": false,
93
+ "galore_target_modules": null,
94
+ "galore_rank": 128,
95
+ "galore_update_proj_gap": 50,
96
+ "galore_scale": 1.0,
97
+ "galore_proj_type": "std",
98
+ "galore_optim_per_parameter": false,
99
+ "galore_with_embedding": false,
100
+ "galore_quantization": false,
101
+ "galore_proj_quant": false,
102
+ "galore_proj_bits": 4,
103
+ "galore_proj_group_size": 256,
104
+ "galore_cos_threshold": 0.4,
105
+ "galore_gamma_proj": 2,
106
+ "galore_queue_size": 5,
107
+ "adalora_target_r": 8,
108
+ "adalora_init_r": 12,
109
+ "adalora_tinit": 0,
110
+ "adalora_tfinal": 0,
111
+ "adalora_deltaT": 1,
112
+ "adalora_beta1": 0.85,
113
+ "adalora_beta2": 0.85,
114
+ "adalora_orth_reg_weight": 0.5,
115
+ "ia3_feedforward_modules": [],
116
+ "llamapro_num_new_blocks": 4,
117
+ "llamapro_num_groups": null,
118
+ "neftune_noise_alpha": null,
119
+ "neftune_backend": "transformers",
120
+ "lisa_activated_layers": 0,
121
+ "lisa_step_interval": 20,
122
+ "reft_layer_key": null,
123
+ "reft_layers": null,
124
+ "reft_rank": 4,
125
+ "reft_intervention_type": "LoreftIntervention",
126
+ "reft_args": null,
127
+ "use_liger": false,
128
+ "gradient_checkpointing": true,
129
+ "deepspeed": null,
130
+ "batch_size": 1,
131
+ "eval_batch_size": 1,
132
+ "auto_find_batch_size": false,
133
+ "num_train_epochs": 2,
134
+ "max_steps": -1,
135
+ "optim": "adamw_torch",
136
+ "adam_beta1": 0.9,
137
+ "adam_beta2": 0.95,
138
+ "adam_epsilon": 1e-08,
139
+ "learning_rate": 0.0001,
140
+ "weight_decay": 0.1,
141
+ "gradient_accumulation_steps": 4,
142
+ "max_grad_norm": 1,
143
+ "predict_with_generate": false,
144
+ "lr_scheduler_type": "cosine",
145
+ "lr_scheduler_kwargs": {},
146
+ "warmup_ratio": 0.05,
147
+ "warmup_steps": 0,
148
+ "eval_steps": 50,
149
+ "save_steps": 50,
150
+ "save_only_model": false,
151
+ "save_total_limit": 2,
152
+ "logging_steps": 5,
153
+ "acc_steps": 1,
154
+ "dataloader_num_workers": 1,
155
+ "dataloader_pin_memory": true,
156
+ "dataloader_drop_last": false,
157
+ "push_to_hub": false,
158
+ "hub_model_id": null,
159
+ "hub_token": null,
160
+ "hub_private_repo": false,
161
+ "hub_strategy": "every_save",
162
+ "test_oom_error": false,
163
+ "disable_tqdm": false,
164
+ "lazy_tokenize": true,
165
+ "preprocess_num_proc": 1,
166
+ "use_flash_attn": null,
167
+ "ignore_args_error": false,
168
+ "check_model_is_latest": true,
169
+ "logging_dir": "/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635/runs",
170
+ "report_to": [
171
+ "tensorboard"
172
+ ],
173
+ "acc_strategy": "token",
174
+ "save_on_each_node": false,
175
+ "evaluation_strategy": "steps",
176
+ "save_strategy": "steps",
177
+ "save_safetensors": true,
178
+ "gpu_memory_fraction": null,
179
+ "include_num_input_tokens_seen": false,
180
+ "local_repo_path": null,
181
+ "custom_register_path": null,
182
+ "custom_dataset_info": null,
183
+ "device_map_config": null,
184
+ "device_max_memory": [],
185
+ "max_new_tokens": 2048,
186
+ "do_sample": null,
187
+ "temperature": null,
188
+ "top_k": null,
189
+ "top_p": null,
190
+ "repetition_penalty": null,
191
+ "num_beams": 1,
192
+ "fsdp": "",
193
+ "fsdp_config": null,
194
+ "sequence_parallel_size": 1,
195
+ "model_layer_cls_name": null,
196
+ "metric_warmup_step": 0,
197
+ "fsdp_num": 1,
198
+ "per_device_train_batch_size": null,
199
+ "per_device_eval_batch_size": null,
200
+ "eval_strategy": null,
201
+ "self_cognition_sample": 0,
202
+ "train_dataset_mix_ratio": 0.0,
203
+ "train_dataset_mix_ds": [
204
+ "ms-bench"
205
+ ],
206
+ "train_dataset_sample": -1,
207
+ "val_dataset_sample": null,
208
+ "safe_serialization": null,
209
+ "only_save_model": null,
210
+ "neftune_alpha": null,
211
+ "deepspeed_config_path": null,
212
+ "model_cache_dir": null,
213
+ "lora_dropout_p": null,
214
+ "lora_target_modules": [],
215
+ "lora_target_regex": null,
216
+ "lora_modules_to_save": [],
217
+ "boft_target_modules": [],
218
+ "boft_modules_to_save": [],
219
+ "vera_target_modules": [],
220
+ "vera_modules_to_save": [],
221
+ "ia3_target_modules": [],
222
+ "ia3_modules_to_save": [],
223
+ "custom_train_dataset_path": [],
224
+ "custom_val_dataset_path": [],
225
+ "device_map_config_path": null,
226
+ "push_hub_strategy": null,
227
+ "use_self_cognition": false,
228
+ "is_multimodal": true,
229
+ "is_vision": true,
230
+ "lora_use_embedding": false,
231
+ "lora_use_all": false,
232
+ "lora_m2s_use_embedding": false,
233
+ "lora_m2s_use_ln": false,
234
+ "torch_dtype": "torch.bfloat16",
235
+ "fp16": false,
236
+ "bf16": true,
237
+ "rank": -1,
238
+ "local_rank": -1,
239
+ "world_size": 1,
240
+ "local_world_size": 1,
241
+ "bnb_4bit_compute_dtype": "torch.bfloat16",
242
+ "load_in_4bit": false,
243
+ "load_in_8bit": false,
244
+ "train_sampler_random": true,
245
+ "train_type": "sft",
246
+ "training_args": "Seq2SeqTrainingArguments(output_dir='/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=4, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0001, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1, num_train_epochs=2, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs={}, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=5, logging_nan_inf_filter=True, save_strategy=<IntervalStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=2, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=50, dataloader_num_workers=1, dataloader_prefetch_factor=None, past_index=-1, run_name='/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed=None, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=False, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy=None, push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=False, include_num_input_tokens_seen=False, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=GenerationConfig {\n \"bos_token_id\": 151643,\n \"do_sample\": true,\n \"eos_token_id\": 151645,\n \"max_new_tokens\": 2048,\n \"pad_token_id\": 151643,\n \"temperature\": 0.01,\n \"top_k\": 1,\n \"top_p\": 0.001\n}\n, acc_strategy='token', loss_name=None, additional_saved_files=[], train_sampler_random=True, metric_warmup_step=0, train_dataset_sample=-1)"
247
+ }
qwen2-vl-7b-instruct/trainer_state.json ADDED
@@ -0,0 +1,1202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.55006742,
3
+ "best_model_checkpoint": "/fs/clip-projects/geoguesser/vlms/qwen/output/qwen2-vl-7b-instruct/v5-20241108-053635/checkpoint-500",
4
+ "epoch": 1.9962616822429906,
5
+ "eval_steps": 50,
6
+ "global_step": 534,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "acc": 0.55171263,
13
+ "epoch": 0.003738317757009346,
14
+ "grad_norm": 1.868323802947998,
15
+ "learning_rate": 3.7037037037037037e-06,
16
+ "loss": 1.850384,
17
+ "memory(GiB)": 31.32,
18
+ "step": 1,
19
+ "train_speed(iter/s)": 0.082547
20
+ },
21
+ {
22
+ "acc": 0.51965243,
23
+ "epoch": 0.018691588785046728,
24
+ "grad_norm": 1.9730364084243774,
25
+ "learning_rate": 1.8518518518518518e-05,
26
+ "loss": 2.06072664,
27
+ "memory(GiB)": 39.03,
28
+ "step": 5,
29
+ "train_speed(iter/s)": 0.11522
30
+ },
31
+ {
32
+ "acc": 0.54611473,
33
+ "epoch": 0.037383177570093455,
34
+ "grad_norm": 1.2351425886154175,
35
+ "learning_rate": 3.7037037037037037e-05,
36
+ "loss": 1.92748413,
37
+ "memory(GiB)": 40.38,
38
+ "step": 10,
39
+ "train_speed(iter/s)": 0.120543
40
+ },
41
+ {
42
+ "acc": 0.53391666,
43
+ "epoch": 0.056074766355140186,
44
+ "grad_norm": 1.3700778484344482,
45
+ "learning_rate": 5.555555555555556e-05,
46
+ "loss": 1.8874958,
47
+ "memory(GiB)": 41.74,
48
+ "step": 15,
49
+ "train_speed(iter/s)": 0.122255
50
+ },
51
+ {
52
+ "acc": 0.53817282,
53
+ "epoch": 0.07476635514018691,
54
+ "grad_norm": 1.5009832382202148,
55
+ "learning_rate": 7.407407407407407e-05,
56
+ "loss": 1.8099781,
57
+ "memory(GiB)": 41.74,
58
+ "step": 20,
59
+ "train_speed(iter/s)": 0.123099
60
+ },
61
+ {
62
+ "acc": 0.55321841,
63
+ "epoch": 0.09345794392523364,
64
+ "grad_norm": 1.3406466245651245,
65
+ "learning_rate": 9.25925925925926e-05,
66
+ "loss": 1.6683075,
67
+ "memory(GiB)": 41.74,
68
+ "step": 25,
69
+ "train_speed(iter/s)": 0.123552
70
+ },
71
+ {
72
+ "acc": 0.55131054,
73
+ "epoch": 0.11214953271028037,
74
+ "grad_norm": 1.2054580450057983,
75
+ "learning_rate": 9.999136119166803e-05,
76
+ "loss": 1.78306332,
77
+ "memory(GiB)": 43.11,
78
+ "step": 30,
79
+ "train_speed(iter/s)": 0.120219
80
+ },
81
+ {
82
+ "acc": 0.54619265,
83
+ "epoch": 0.1308411214953271,
84
+ "grad_norm": 1.3058720827102661,
85
+ "learning_rate": 9.99385792841537e-05,
86
+ "loss": 1.81088448,
87
+ "memory(GiB)": 43.11,
88
+ "step": 35,
89
+ "train_speed(iter/s)": 0.120961
90
+ },
91
+ {
92
+ "acc": 0.56460981,
93
+ "epoch": 0.14953271028037382,
94
+ "grad_norm": 1.2871434688568115,
95
+ "learning_rate": 9.983786540671051e-05,
96
+ "loss": 1.69376984,
97
+ "memory(GiB)": 44.48,
98
+ "step": 40,
99
+ "train_speed(iter/s)": 0.121524
100
+ },
101
+ {
102
+ "acc": 0.57231364,
103
+ "epoch": 0.16822429906542055,
104
+ "grad_norm": 1.1757748126983643,
105
+ "learning_rate": 9.968931622637652e-05,
106
+ "loss": 1.6526125,
107
+ "memory(GiB)": 44.48,
108
+ "step": 45,
109
+ "train_speed(iter/s)": 0.121972
110
+ },
111
+ {
112
+ "acc": 0.56770124,
113
+ "epoch": 0.18691588785046728,
114
+ "grad_norm": 1.3221774101257324,
115
+ "learning_rate": 9.949307432339625e-05,
116
+ "loss": 1.70850391,
117
+ "memory(GiB)": 44.48,
118
+ "step": 50,
119
+ "train_speed(iter/s)": 0.122298
120
+ },
121
+ {
122
+ "epoch": 0.18691588785046728,
123
+ "eval_acc": 0.5727995201679412,
124
+ "eval_loss": 1.6454861164093018,
125
+ "eval_runtime": 60.9474,
126
+ "eval_samples_per_second": 0.82,
127
+ "eval_steps_per_second": 0.82,
128
+ "step": 50
129
+ },
130
+ {
131
+ "acc": 0.56792145,
132
+ "epoch": 0.205607476635514,
133
+ "grad_norm": 1.3802762031555176,
134
+ "learning_rate": 9.924932805436949e-05,
135
+ "loss": 1.61318073,
136
+ "memory(GiB)": 44.48,
137
+ "step": 55,
138
+ "train_speed(iter/s)": 0.10774
139
+ },
140
+ {
141
+ "acc": 0.56201911,
142
+ "epoch": 0.22429906542056074,
143
+ "grad_norm": 1.4624619483947754,
144
+ "learning_rate": 9.895831137146318e-05,
145
+ "loss": 1.68176594,
146
+ "memory(GiB)": 44.48,
147
+ "step": 60,
148
+ "train_speed(iter/s)": 0.109037
149
+ },
150
+ {
151
+ "acc": 0.56515856,
152
+ "epoch": 0.24299065420560748,
153
+ "grad_norm": 1.3205868005752563,
154
+ "learning_rate": 9.862030359785981e-05,
155
+ "loss": 1.65190887,
156
+ "memory(GiB)": 44.48,
157
+ "step": 65,
158
+ "train_speed(iter/s)": 0.110146
159
+ },
160
+ {
161
+ "acc": 0.55511956,
162
+ "epoch": 0.2616822429906542,
163
+ "grad_norm": 1.3879112005233765,
164
+ "learning_rate": 9.82356291596578e-05,
165
+ "loss": 1.6682188,
166
+ "memory(GiB)": 44.48,
167
+ "step": 70,
168
+ "train_speed(iter/s)": 0.111111
169
+ },
170
+ {
171
+ "acc": 0.56104274,
172
+ "epoch": 0.2803738317757009,
173
+ "grad_norm": 1.3009270429611206,
174
+ "learning_rate": 9.780465727448149e-05,
175
+ "loss": 1.7461134,
176
+ "memory(GiB)": 45.86,
177
+ "step": 75,
178
+ "train_speed(iter/s)": 0.110878
179
+ },
180
+ {
181
+ "acc": 0.57672982,
182
+ "epoch": 0.29906542056074764,
183
+ "grad_norm": 1.3524978160858154,
184
+ "learning_rate": 9.732780159709912e-05,
185
+ "loss": 1.61342182,
186
+ "memory(GiB)": 45.86,
187
+ "step": 80,
188
+ "train_speed(iter/s)": 0.111688
189
+ },
190
+ {
191
+ "acc": 0.56858454,
192
+ "epoch": 0.3177570093457944,
193
+ "grad_norm": 1.368619680404663,
194
+ "learning_rate": 9.680551982238942e-05,
195
+ "loss": 1.62513695,
196
+ "memory(GiB)": 45.86,
197
+ "step": 85,
198
+ "train_speed(iter/s)": 0.112415
199
+ },
200
+ {
201
+ "acc": 0.56374822,
202
+ "epoch": 0.3364485981308411,
203
+ "grad_norm": 1.397831916809082,
204
+ "learning_rate": 9.623831324603754e-05,
205
+ "loss": 1.69306774,
206
+ "memory(GiB)": 45.86,
207
+ "step": 90,
208
+ "train_speed(iter/s)": 0.113063
209
+ },
210
+ {
211
+ "acc": 0.57811651,
212
+ "epoch": 0.35514018691588783,
213
+ "grad_norm": 1.271440029144287,
214
+ "learning_rate": 9.562672628338233e-05,
215
+ "loss": 1.63228798,
216
+ "memory(GiB)": 45.86,
217
+ "step": 95,
218
+ "train_speed(iter/s)": 0.113645
219
+ },
220
+ {
221
+ "acc": 0.5570353,
222
+ "epoch": 0.37383177570093457,
223
+ "grad_norm": 1.2964327335357666,
224
+ "learning_rate": 9.497134594687634e-05,
225
+ "loss": 1.72664585,
226
+ "memory(GiB)": 45.86,
227
+ "step": 100,
228
+ "train_speed(iter/s)": 0.114174
229
+ },
230
+ {
231
+ "epoch": 0.37383177570093457,
232
+ "eval_acc": 0.5792472634577898,
233
+ "eval_loss": 1.6085342168807983,
234
+ "eval_runtime": 62.4797,
235
+ "eval_samples_per_second": 0.8,
236
+ "eval_steps_per_second": 0.8,
237
+ "step": 100
238
+ },
239
+ {
240
+ "acc": 0.57630959,
241
+ "epoch": 0.3925233644859813,
242
+ "grad_norm": 1.244130253791809,
243
+ "learning_rate": 9.42728012826605e-05,
244
+ "loss": 1.64715214,
245
+ "memory(GiB)": 45.86,
246
+ "step": 105,
247
+ "train_speed(iter/s)": 0.107229
248
+ },
249
+ {
250
+ "acc": 0.5584549,
251
+ "epoch": 0.411214953271028,
252
+ "grad_norm": 1.3243989944458008,
253
+ "learning_rate": 9.353176276679396e-05,
254
+ "loss": 1.68698692,
255
+ "memory(GiB)": 45.86,
256
+ "step": 110,
257
+ "train_speed(iter/s)": 0.107951
258
+ },
259
+ {
260
+ "acc": 0.5546257,
261
+ "epoch": 0.42990654205607476,
262
+ "grad_norm": 1.437445878982544,
263
+ "learning_rate": 9.274894166171888e-05,
264
+ "loss": 1.66922894,
265
+ "memory(GiB)": 45.86,
266
+ "step": 115,
267
+ "train_speed(iter/s)": 0.10862
268
+ },
269
+ {
270
+ "acc": 0.57244515,
271
+ "epoch": 0.4485981308411215,
272
+ "grad_norm": 1.3543046712875366,
273
+ "learning_rate": 9.192508933357753e-05,
274
+ "loss": 1.70311775,
275
+ "memory(GiB)": 45.86,
276
+ "step": 120,
277
+ "train_speed(iter/s)": 0.109232
278
+ },
279
+ {
280
+ "acc": 0.56850109,
281
+ "epoch": 0.4672897196261682,
282
+ "grad_norm": 1.287984013557434,
283
+ "learning_rate": 9.106099653103728e-05,
284
+ "loss": 1.61406059,
285
+ "memory(GiB)": 45.86,
286
+ "step": 125,
287
+ "train_speed(iter/s)": 0.109801
288
+ },
289
+ {
290
+ "acc": 0.56755419,
291
+ "epoch": 0.48598130841121495,
292
+ "grad_norm": 1.4639618396759033,
293
+ "learning_rate": 9.015749262631536e-05,
294
+ "loss": 1.57637978,
295
+ "memory(GiB)": 45.86,
296
+ "step": 130,
297
+ "train_speed(iter/s)": 0.110329
298
+ },
299
+ {
300
+ "acc": 0.58115373,
301
+ "epoch": 0.5046728971962616,
302
+ "grad_norm": 1.5570566654205322,
303
+ "learning_rate": 8.921544481913218e-05,
304
+ "loss": 1.62401295,
305
+ "memory(GiB)": 45.86,
306
+ "step": 135,
307
+ "train_speed(iter/s)": 0.110827
308
+ },
309
+ {
310
+ "acc": 0.55897279,
311
+ "epoch": 0.5233644859813084,
312
+ "grad_norm": 1.4730037450790405,
313
+ "learning_rate": 8.823575730435693e-05,
314
+ "loss": 1.66579857,
315
+ "memory(GiB)": 52.51,
316
+ "step": 140,
317
+ "train_speed(iter/s)": 0.111291
318
+ },
319
+ {
320
+ "acc": 0.56799178,
321
+ "epoch": 0.5420560747663551,
322
+ "grad_norm": 1.350874423980713,
323
+ "learning_rate": 8.721937040414481e-05,
324
+ "loss": 1.60019073,
325
+ "memory(GiB)": 52.51,
326
+ "step": 145,
327
+ "train_speed(iter/s)": 0.111724
328
+ },
329
+ {
330
+ "acc": 0.55238876,
331
+ "epoch": 0.5607476635514018,
332
+ "grad_norm": 1.5056456327438354,
333
+ "learning_rate": 8.616725966539832e-05,
334
+ "loss": 1.68097,
335
+ "memory(GiB)": 52.51,
336
+ "step": 150,
337
+ "train_speed(iter/s)": 0.11205
338
+ },
339
+ {
340
+ "epoch": 0.5607476635514018,
341
+ "eval_acc": 0.5831458989353726,
342
+ "eval_loss": 1.588950753211975,
343
+ "eval_runtime": 60.5954,
344
+ "eval_samples_per_second": 0.825,
345
+ "eval_steps_per_second": 0.825,
346
+ "step": 150
347
+ },
348
+ {
349
+ "acc": 0.56648855,
350
+ "epoch": 0.5794392523364486,
351
+ "grad_norm": 1.4353731870651245,
352
+ "learning_rate": 8.508043492341944e-05,
353
+ "loss": 1.61546593,
354
+ "memory(GiB)": 52.51,
355
+ "step": 155,
356
+ "train_speed(iter/s)": 0.107639
357
+ },
358
+ {
359
+ "acc": 0.57423716,
360
+ "epoch": 0.5981308411214953,
361
+ "grad_norm": 1.514600396156311,
362
+ "learning_rate": 8.395993933265101e-05,
363
+ "loss": 1.65116329,
364
+ "memory(GiB)": 52.51,
365
+ "step": 160,
366
+ "train_speed(iter/s)": 0.108123
367
+ },
368
+ {
369
+ "acc": 0.56541142,
370
+ "epoch": 0.616822429906542,
371
+ "grad_norm": 1.3241384029388428,
372
+ "learning_rate": 8.280684836543794e-05,
373
+ "loss": 1.65839729,
374
+ "memory(GiB)": 52.51,
375
+ "step": 165,
376
+ "train_speed(iter/s)": 0.108573
377
+ },
378
+ {
379
+ "acc": 0.57026463,
380
+ "epoch": 0.6355140186915887,
381
+ "grad_norm": 1.3388739824295044,
382
+ "learning_rate": 8.162226877976887e-05,
383
+ "loss": 1.61102333,
384
+ "memory(GiB)": 52.51,
385
+ "step": 170,
386
+ "train_speed(iter/s)": 0.109007
387
+ },
388
+ {
389
+ "acc": 0.57890859,
390
+ "epoch": 0.6542056074766355,
391
+ "grad_norm": 1.37869131565094,
392
+ "learning_rate": 8.040733755698955e-05,
393
+ "loss": 1.60712547,
394
+ "memory(GiB)": 52.51,
395
+ "step": 175,
396
+ "train_speed(iter/s)": 0.109415
397
+ },
398
+ {
399
+ "acc": 0.57019186,
400
+ "epoch": 0.6728971962616822,
401
+ "grad_norm": 1.4313998222351074,
402
+ "learning_rate": 7.916322081050709e-05,
403
+ "loss": 1.62115898,
404
+ "memory(GiB)": 52.51,
405
+ "step": 180,
406
+ "train_speed(iter/s)": 0.109805
407
+ },
408
+ {
409
+ "acc": 0.57807865,
410
+ "epoch": 0.6915887850467289,
411
+ "grad_norm": 1.3123388290405273,
412
+ "learning_rate": 7.789111266653285e-05,
413
+ "loss": 1.63029137,
414
+ "memory(GiB)": 52.51,
415
+ "step": 185,
416
+ "train_speed(iter/s)": 0.110173
417
+ },
418
+ {
419
+ "acc": 0.58090611,
420
+ "epoch": 0.7102803738317757,
421
+ "grad_norm": 1.460463047027588,
422
+ "learning_rate": 7.659223411793798e-05,
423
+ "loss": 1.57071505,
424
+ "memory(GiB)": 52.51,
425
+ "step": 190,
426
+ "train_speed(iter/s)": 0.110531
427
+ },
428
+ {
429
+ "acc": 0.57307801,
430
+ "epoch": 0.7289719626168224,
431
+ "grad_norm": 1.3995453119277954,
432
+ "learning_rate": 7.526783185232207e-05,
433
+ "loss": 1.61080112,
434
+ "memory(GiB)": 52.51,
435
+ "step": 195,
436
+ "train_speed(iter/s)": 0.110867
437
+ },
438
+ {
439
+ "acc": 0.5799108,
440
+ "epoch": 0.7476635514018691,
441
+ "grad_norm": 1.4361484050750732,
442
+ "learning_rate": 7.391917705541927e-05,
443
+ "loss": 1.64733868,
444
+ "memory(GiB)": 52.51,
445
+ "step": 200,
446
+ "train_speed(iter/s)": 0.111188
447
+ },
448
+ {
449
+ "epoch": 0.7476635514018691,
450
+ "eval_acc": 0.5834457939721097,
451
+ "eval_loss": 1.570568561553955,
452
+ "eval_runtime": 60.5903,
453
+ "eval_samples_per_second": 0.825,
454
+ "eval_steps_per_second": 0.825,
455
+ "step": 200
456
+ },
457
+ {
458
+ "acc": 0.56698923,
459
+ "epoch": 0.7663551401869159,
460
+ "grad_norm": 1.3287904262542725,
461
+ "learning_rate": 7.254756419099074e-05,
462
+ "loss": 1.64705162,
463
+ "memory(GiB)": 52.51,
464
+ "step": 205,
465
+ "train_speed(iter/s)": 0.107887
466
+ },
467
+ {
468
+ "acc": 0.57151198,
469
+ "epoch": 0.7850467289719626,
470
+ "grad_norm": 1.38331139087677,
471
+ "learning_rate": 7.115430975837457e-05,
472
+ "loss": 1.64652443,
473
+ "memory(GiB)": 52.51,
474
+ "step": 210,
475
+ "train_speed(iter/s)": 0.108252
476
+ },
477
+ {
478
+ "acc": 0.58841505,
479
+ "epoch": 0.8037383177570093,
480
+ "grad_norm": 1.5937939882278442,
481
+ "learning_rate": 6.974075102888536e-05,
482
+ "loss": 1.61707039,
483
+ "memory(GiB)": 52.51,
484
+ "step": 215,
485
+ "train_speed(iter/s)": 0.108603
486
+ },
487
+ {
488
+ "acc": 0.55511918,
489
+ "epoch": 0.822429906542056,
490
+ "grad_norm": 1.6487551927566528,
491
+ "learning_rate": 6.830824476227646e-05,
492
+ "loss": 1.65553608,
493
+ "memory(GiB)": 52.51,
494
+ "step": 220,
495
+ "train_speed(iter/s)": 0.108935
496
+ },
497
+ {
498
+ "acc": 0.58533549,
499
+ "epoch": 0.8411214953271028,
500
+ "grad_norm": 1.4343266487121582,
501
+ "learning_rate": 6.685816590449708e-05,
502
+ "loss": 1.58468885,
503
+ "memory(GiB)": 52.51,
504
+ "step": 225,
505
+ "train_speed(iter/s)": 0.109256
506
+ },
507
+ {
508
+ "acc": 0.57694592,
509
+ "epoch": 0.8598130841121495,
510
+ "grad_norm": 1.368004560470581,
511
+ "learning_rate": 6.539190626799366e-05,
512
+ "loss": 1.60840836,
513
+ "memory(GiB)": 52.51,
514
+ "step": 230,
515
+ "train_speed(iter/s)": 0.109563
516
+ },
517
+ {
518
+ "acc": 0.57554379,
519
+ "epoch": 0.8785046728971962,
520
+ "grad_norm": 1.513482928276062,
521
+ "learning_rate": 6.391087319582264e-05,
522
+ "loss": 1.59513159,
523
+ "memory(GiB)": 52.51,
524
+ "step": 235,
525
+ "train_speed(iter/s)": 0.109855
526
+ },
527
+ {
528
+ "acc": 0.56200686,
529
+ "epoch": 0.897196261682243,
530
+ "grad_norm": 1.447696566581726,
531
+ "learning_rate": 6.241648821085666e-05,
532
+ "loss": 1.61208496,
533
+ "memory(GiB)": 52.51,
534
+ "step": 240,
535
+ "train_speed(iter/s)": 0.110135
536
+ },
537
+ {
538
+ "acc": 0.57686815,
539
+ "epoch": 0.9158878504672897,
540
+ "grad_norm": 1.4834848642349243,
541
+ "learning_rate": 6.0910185651380626e-05,
542
+ "loss": 1.56525345,
543
+ "memory(GiB)": 52.51,
544
+ "step": 245,
545
+ "train_speed(iter/s)": 0.110415
546
+ },
547
+ {
548
+ "acc": 0.57838049,
549
+ "epoch": 0.9345794392523364,
550
+ "grad_norm": 1.4449986219406128,
551
+ "learning_rate": 5.939341129438739e-05,
552
+ "loss": 1.66088371,
553
+ "memory(GiB)": 52.51,
554
+ "step": 250,
555
+ "train_speed(iter/s)": 0.110677
556
+ },
557
+ {
558
+ "epoch": 0.9345794392523364,
559
+ "eval_acc": 0.5871944819313241,
560
+ "eval_loss": 1.5589616298675537,
561
+ "eval_runtime": 60.6063,
562
+ "eval_samples_per_second": 0.825,
563
+ "eval_steps_per_second": 0.825,
564
+ "step": 250
565
+ },
566
+ {
567
+ "acc": 0.56091037,
568
+ "epoch": 0.9532710280373832,
569
+ "grad_norm": 1.4498945474624634,
570
+ "learning_rate": 5.786762096789431e-05,
571
+ "loss": 1.6876915,
572
+ "memory(GiB)": 52.51,
573
+ "step": 255,
574
+ "train_speed(iter/s)": 0.108045
575
+ },
576
+ {
577
+ "acc": 0.57053814,
578
+ "epoch": 0.9719626168224299,
579
+ "grad_norm": 1.2757234573364258,
580
+ "learning_rate": 5.633427915361261e-05,
581
+ "loss": 1.65799484,
582
+ "memory(GiB)": 52.51,
583
+ "step": 260,
584
+ "train_speed(iter/s)": 0.108333
585
+ },
586
+ {
587
+ "acc": 0.56272326,
588
+ "epoch": 0.9906542056074766,
589
+ "grad_norm": 1.4214109182357788,
590
+ "learning_rate": 5.479485758131089e-05,
591
+ "loss": 1.64700985,
592
+ "memory(GiB)": 52.51,
593
+ "step": 265,
594
+ "train_speed(iter/s)": 0.108614
595
+ },
596
+ {
597
+ "acc": 0.59196057,
598
+ "epoch": 1.0093457943925233,
599
+ "grad_norm": 1.255962610244751,
600
+ "learning_rate": 5.325083381622165e-05,
601
+ "loss": 1.56780367,
602
+ "memory(GiB)": 52.51,
603
+ "step": 270,
604
+ "train_speed(iter/s)": 0.108871
605
+ },
606
+ {
607
+ "acc": 0.60109649,
608
+ "epoch": 1.02803738317757,
609
+ "grad_norm": 1.4240363836288452,
610
+ "learning_rate": 5.1703689840846945e-05,
611
+ "loss": 1.45532875,
612
+ "memory(GiB)": 52.51,
613
+ "step": 275,
614
+ "train_speed(iter/s)": 0.109138
615
+ },
616
+ {
617
+ "acc": 0.59727616,
618
+ "epoch": 1.0467289719626167,
619
+ "grad_norm": 1.5935661792755127,
620
+ "learning_rate": 5.01549106325243e-05,
621
+ "loss": 1.51683445,
622
+ "memory(GiB)": 52.51,
623
+ "step": 280,
624
+ "train_speed(iter/s)": 0.109392
625
+ },
626
+ {
627
+ "acc": 0.62937155,
628
+ "epoch": 1.0654205607476634,
629
+ "grad_norm": 1.6722455024719238,
630
+ "learning_rate": 4.860598273811792e-05,
631
+ "loss": 1.35466251,
632
+ "memory(GiB)": 52.51,
633
+ "step": 285,
634
+ "train_speed(iter/s)": 0.109642
635
+ },
636
+ {
637
+ "acc": 0.58850698,
638
+ "epoch": 1.0841121495327102,
639
+ "grad_norm": 1.524778127670288,
640
+ "learning_rate": 4.705839284720376e-05,
641
+ "loss": 1.48758812,
642
+ "memory(GiB)": 52.51,
643
+ "step": 290,
644
+ "train_speed(iter/s)": 0.109812
645
+ },
646
+ {
647
+ "acc": 0.60075417,
648
+ "epoch": 1.102803738317757,
649
+ "grad_norm": 1.757370114326477,
650
+ "learning_rate": 4.55136263651172e-05,
651
+ "loss": 1.50896826,
652
+ "memory(GiB)": 52.51,
653
+ "step": 295,
654
+ "train_speed(iter/s)": 0.110044
655
+ },
656
+ {
657
+ "acc": 0.61744561,
658
+ "epoch": 1.1214953271028036,
659
+ "grad_norm": 2.0011301040649414,
660
+ "learning_rate": 4.397316598723385e-05,
661
+ "loss": 1.42747393,
662
+ "memory(GiB)": 52.51,
663
+ "step": 300,
664
+ "train_speed(iter/s)": 0.110269
665
+ },
666
+ {
667
+ "epoch": 1.1214953271028036,
668
+ "eval_acc": 0.5852451641925326,
669
+ "eval_loss": 1.5581213235855103,
670
+ "eval_runtime": 60.6131,
671
+ "eval_samples_per_second": 0.825,
672
+ "eval_steps_per_second": 0.825,
673
+ "step": 300
674
+ },
675
+ {
676
+ "acc": 0.59082665,
677
+ "epoch": 1.1401869158878504,
678
+ "grad_norm": 1.8005529642105103,
679
+ "learning_rate": 4.243849027585096e-05,
680
+ "loss": 1.49810066,
681
+ "memory(GiB)": 52.51,
682
+ "step": 305,
683
+ "train_speed(iter/s)": 0.108082
684
+ },
685
+ {
686
+ "acc": 0.58803234,
687
+ "epoch": 1.158878504672897,
688
+ "grad_norm": 1.8836215734481812,
689
+ "learning_rate": 4.0911072241036194e-05,
690
+ "loss": 1.53769073,
691
+ "memory(GiB)": 52.51,
692
+ "step": 310,
693
+ "train_speed(iter/s)": 0.108325
694
+ },
695
+ {
696
+ "acc": 0.61663084,
697
+ "epoch": 1.1775700934579438,
698
+ "grad_norm": 1.7952263355255127,
699
+ "learning_rate": 3.9392377926805226e-05,
700
+ "loss": 1.44214535,
701
+ "memory(GiB)": 52.51,
702
+ "step": 315,
703
+ "train_speed(iter/s)": 0.108566
704
+ },
705
+ {
706
+ "acc": 0.59271388,
707
+ "epoch": 1.1962616822429906,
708
+ "grad_norm": 1.8852580785751343,
709
+ "learning_rate": 3.788386500398583e-05,
710
+ "loss": 1.49927893,
711
+ "memory(GiB)": 52.51,
712
+ "step": 320,
713
+ "train_speed(iter/s)": 0.108797
714
+ },
715
+ {
716
+ "acc": 0.60236468,
717
+ "epoch": 1.2149532710280373,
718
+ "grad_norm": 1.737602949142456,
719
+ "learning_rate": 3.6386981371118355e-05,
720
+ "loss": 1.42521906,
721
+ "memory(GiB)": 52.51,
722
+ "step": 325,
723
+ "train_speed(iter/s)": 0.109019
724
+ },
725
+ {
726
+ "acc": 0.6055068,
727
+ "epoch": 1.233644859813084,
728
+ "grad_norm": 1.914955496788025,
729
+ "learning_rate": 3.49031637647361e-05,
730
+ "loss": 1.47248116,
731
+ "memory(GiB)": 52.51,
732
+ "step": 330,
733
+ "train_speed(iter/s)": 0.109238
734
+ },
735
+ {
736
+ "acc": 0.61518903,
737
+ "epoch": 1.2523364485981308,
738
+ "grad_norm": 1.7206995487213135,
739
+ "learning_rate": 3.343383638035902e-05,
740
+ "loss": 1.38390493,
741
+ "memory(GiB)": 52.51,
742
+ "step": 335,
743
+ "train_speed(iter/s)": 0.109447
744
+ },
745
+ {
746
+ "acc": 0.60801978,
747
+ "epoch": 1.2710280373831775,
748
+ "grad_norm": 1.9262409210205078,
749
+ "learning_rate": 3.1980409505524544e-05,
750
+ "loss": 1.41381416,
751
+ "memory(GiB)": 52.51,
752
+ "step": 340,
753
+ "train_speed(iter/s)": 0.109652
754
+ },
755
+ {
756
+ "acc": 0.60384398,
757
+ "epoch": 1.2897196261682242,
758
+ "grad_norm": 2.144967794418335,
759
+ "learning_rate": 3.054427816616773e-05,
760
+ "loss": 1.40025005,
761
+ "memory(GiB)": 52.51,
762
+ "step": 345,
763
+ "train_speed(iter/s)": 0.109855
764
+ },
765
+ {
766
+ "acc": 0.60187116,
767
+ "epoch": 1.308411214953271,
768
+ "grad_norm": 2.0876433849334717,
769
+ "learning_rate": 2.91268207876494e-05,
770
+ "loss": 1.44376688,
771
+ "memory(GiB)": 52.51,
772
+ "step": 350,
773
+ "train_speed(iter/s)": 0.110051
774
+ },
775
+ {
776
+ "epoch": 1.308411214953271,
777
+ "eval_acc": 0.5856950067476383,
778
+ "eval_loss": 1.5565516948699951,
779
+ "eval_runtime": 60.5775,
780
+ "eval_samples_per_second": 0.825,
781
+ "eval_steps_per_second": 0.825,
782
+ "step": 350
783
+ },
784
+ {
785
+ "acc": 0.61420636,
786
+ "epoch": 1.3271028037383177,
787
+ "grad_norm": 1.9940565824508667,
788
+ "learning_rate": 2.7729397871718304e-05,
789
+ "loss": 1.40987692,
790
+ "memory(GiB)": 52.51,
791
+ "step": 355,
792
+ "train_speed(iter/s)": 0.108178
793
+ },
794
+ {
795
+ "acc": 0.59907641,
796
+ "epoch": 1.3457943925233644,
797
+ "grad_norm": 1.9915255308151245,
798
+ "learning_rate": 2.635335069067617e-05,
799
+ "loss": 1.44835072,
800
+ "memory(GiB)": 52.51,
801
+ "step": 360,
802
+ "train_speed(iter/s)": 0.108387
803
+ },
804
+ {
805
+ "acc": 0.62042379,
806
+ "epoch": 1.3644859813084111,
807
+ "grad_norm": 2.130258798599243,
808
+ "learning_rate": 2.500000000000001e-05,
809
+ "loss": 1.41162367,
810
+ "memory(GiB)": 52.51,
811
+ "step": 365,
812
+ "train_speed(iter/s)": 0.108589
813
+ },
814
+ {
815
+ "acc": 0.6139565,
816
+ "epoch": 1.3831775700934579,
817
+ "grad_norm": 1.8815335035324097,
818
+ "learning_rate": 2.367064477065652e-05,
819
+ "loss": 1.41061649,
820
+ "memory(GiB)": 52.51,
821
+ "step": 370,
822
+ "train_speed(iter/s)": 0.108788
823
+ },
824
+ {
825
+ "acc": 0.60995245,
826
+ "epoch": 1.4018691588785046,
827
+ "grad_norm": 2.237551689147949,
828
+ "learning_rate": 2.2366560942325832e-05,
829
+ "loss": 1.41165752,
830
+ "memory(GiB)": 52.51,
831
+ "step": 375,
832
+ "train_speed(iter/s)": 0.108982
833
+ },
834
+ {
835
+ "acc": 0.61310611,
836
+ "epoch": 1.4205607476635513,
837
+ "grad_norm": 2.2738187313079834,
838
+ "learning_rate": 2.108900019873103e-05,
839
+ "loss": 1.46329918,
840
+ "memory(GiB)": 52.51,
841
+ "step": 380,
842
+ "train_speed(iter/s)": 0.109169
843
+ },
844
+ {
845
+ "acc": 0.61972389,
846
+ "epoch": 1.439252336448598,
847
+ "grad_norm": 2.050431966781616,
848
+ "learning_rate": 1.983918876624902e-05,
849
+ "loss": 1.39380827,
850
+ "memory(GiB)": 52.51,
851
+ "step": 385,
852
+ "train_speed(iter/s)": 0.109353
853
+ },
854
+ {
855
+ "acc": 0.60818005,
856
+ "epoch": 1.4579439252336448,
857
+ "grad_norm": 2.2794229984283447,
858
+ "learning_rate": 1.8618326236955907e-05,
859
+ "loss": 1.46415033,
860
+ "memory(GiB)": 52.51,
861
+ "step": 390,
862
+ "train_speed(iter/s)": 0.109532
863
+ },
864
+ {
865
+ "acc": 0.59707479,
866
+ "epoch": 1.4766355140186915,
867
+ "grad_norm": 2.2006595134735107,
868
+ "learning_rate": 1.7427584417236194e-05,
869
+ "loss": 1.49114666,
870
+ "memory(GiB)": 52.51,
871
+ "step": 395,
872
+ "train_speed(iter/s)": 0.109705
873
+ },
874
+ {
875
+ "acc": 0.6111486,
876
+ "epoch": 1.4953271028037383,
877
+ "grad_norm": 2.0496108531951904,
878
+ "learning_rate": 1.626810620306163e-05,
879
+ "loss": 1.38812447,
880
+ "memory(GiB)": 52.51,
881
+ "step": 400,
882
+ "train_speed(iter/s)": 0.109877
883
+ },
884
+ {
885
+ "epoch": 1.4953271028037383,
886
+ "eval_acc": 0.5873444294496926,
887
+ "eval_loss": 1.5544381141662598,
888
+ "eval_runtime": 60.5613,
889
+ "eval_samples_per_second": 0.826,
890
+ "eval_steps_per_second": 0.826,
891
+ "step": 400
892
+ },
893
+ {
894
+ "acc": 0.60279655,
895
+ "epoch": 1.514018691588785,
896
+ "grad_norm": 1.954108476638794,
897
+ "learning_rate": 1.5141004483018323e-05,
898
+ "loss": 1.44826994,
899
+ "memory(GiB)": 52.51,
900
+ "step": 405,
901
+ "train_speed(iter/s)": 0.108237
902
+ },
903
+ {
904
+ "acc": 0.60491271,
905
+ "epoch": 1.5327102803738317,
906
+ "grad_norm": 2.4498937129974365,
907
+ "learning_rate": 1.4047361070135995e-05,
908
+ "loss": 1.4636652,
909
+ "memory(GiB)": 52.51,
910
+ "step": 410,
911
+ "train_speed(iter/s)": 0.108423
912
+ },
913
+ {
914
+ "acc": 0.59805059,
915
+ "epoch": 1.5514018691588785,
916
+ "grad_norm": 1.9891496896743774,
917
+ "learning_rate": 1.2988225663543602e-05,
918
+ "loss": 1.51361618,
919
+ "memory(GiB)": 52.51,
920
+ "step": 415,
921
+ "train_speed(iter/s)": 0.108601
922
+ },
923
+ {
924
+ "acc": 0.60604153,
925
+ "epoch": 1.5700934579439252,
926
+ "grad_norm": 2.281243324279785,
927
+ "learning_rate": 1.1964614840949002e-05,
928
+ "loss": 1.43464155,
929
+ "memory(GiB)": 52.51,
930
+ "step": 420,
931
+ "train_speed(iter/s)": 0.108777
932
+ },
933
+ {
934
+ "acc": 0.59663863,
935
+ "epoch": 1.588785046728972,
936
+ "grad_norm": 2.1692161560058594,
937
+ "learning_rate": 1.097751108290867e-05,
938
+ "loss": 1.47755518,
939
+ "memory(GiB)": 52.51,
940
+ "step": 425,
941
+ "train_speed(iter/s)": 0.108947
942
+ },
943
+ {
944
+ "acc": 0.62566915,
945
+ "epoch": 1.6074766355140186,
946
+ "grad_norm": 2.370448112487793,
947
+ "learning_rate": 1.0027861829824952e-05,
948
+ "loss": 1.36240664,
949
+ "memory(GiB)": 52.51,
950
+ "step": 430,
951
+ "train_speed(iter/s)": 0.109117
952
+ },
953
+ {
954
+ "acc": 0.60366473,
955
+ "epoch": 1.6261682242990654,
956
+ "grad_norm": 2.143240451812744,
957
+ "learning_rate": 9.11657857257509e-06,
958
+ "loss": 1.49398394,
959
+ "memory(GiB)": 52.51,
960
+ "step": 435,
961
+ "train_speed(iter/s)": 0.109226
962
+ },
963
+ {
964
+ "acc": 0.60729022,
965
+ "epoch": 1.644859813084112,
966
+ "grad_norm": 2.266324758529663,
967
+ "learning_rate": 8.244535977645585e-06,
968
+ "loss": 1.4582058,
969
+ "memory(GiB)": 52.51,
970
+ "step": 440,
971
+ "train_speed(iter/s)": 0.109388
972
+ },
973
+ {
974
+ "acc": 0.61000395,
975
+ "epoch": 1.6635514018691588,
976
+ "grad_norm": 2.243384599685669,
977
+ "learning_rate": 7.412571047611155e-06,
978
+ "loss": 1.39406261,
979
+ "memory(GiB)": 52.51,
980
+ "step": 445,
981
+ "train_speed(iter/s)": 0.109547
982
+ },
983
+ {
984
+ "acc": 0.60550241,
985
+ "epoch": 1.6822429906542056,
986
+ "grad_norm": 2.3402411937713623,
987
+ "learning_rate": 6.621482317764105e-06,
988
+ "loss": 1.44629755,
989
+ "memory(GiB)": 52.51,
990
+ "step": 450,
991
+ "train_speed(iter/s)": 0.109702
992
+ },
993
+ {
994
+ "epoch": 1.6822429906542056,
995
+ "eval_acc": 0.5838956365272154,
996
+ "eval_loss": 1.5518497228622437,
997
+ "eval_runtime": 60.463,
998
+ "eval_samples_per_second": 0.827,
999
+ "eval_steps_per_second": 0.827,
1000
+ "step": 450
1001
+ },
1002
+ {
1003
+ "acc": 0.62354083,
1004
+ "epoch": 1.7009345794392523,
1005
+ "grad_norm": 2.3499748706817627,
1006
+ "learning_rate": 5.872029089665587e-06,
1007
+ "loss": 1.36534414,
1008
+ "memory(GiB)": 52.51,
1009
+ "step": 455,
1010
+ "train_speed(iter/s)": 0.108251
1011
+ },
1012
+ {
1013
+ "acc": 0.60730128,
1014
+ "epoch": 1.719626168224299,
1015
+ "grad_norm": 2.479720115661621,
1016
+ "learning_rate": 5.164930702353782e-06,
1017
+ "loss": 1.44677553,
1018
+ "memory(GiB)": 52.51,
1019
+ "step": 460,
1020
+ "train_speed(iter/s)": 0.108417
1021
+ },
1022
+ {
1023
+ "acc": 0.59804258,
1024
+ "epoch": 1.7383177570093458,
1025
+ "grad_norm": 2.117152214050293,
1026
+ "learning_rate": 4.500865841909168e-06,
1027
+ "loss": 1.46659861,
1028
+ "memory(GiB)": 52.51,
1029
+ "step": 465,
1030
+ "train_speed(iter/s)": 0.108577
1031
+ },
1032
+ {
1033
+ "acc": 0.60334945,
1034
+ "epoch": 1.7570093457943925,
1035
+ "grad_norm": 2.2500483989715576,
1036
+ "learning_rate": 3.880471890038967e-06,
1037
+ "loss": 1.4467123,
1038
+ "memory(GiB)": 52.51,
1039
+ "step": 470,
1040
+ "train_speed(iter/s)": 0.108736
1041
+ },
1042
+ {
1043
+ "acc": 0.60877209,
1044
+ "epoch": 1.7757009345794392,
1045
+ "grad_norm": 2.166339635848999,
1046
+ "learning_rate": 3.3043443123065286e-06,
1047
+ "loss": 1.49398079,
1048
+ "memory(GiB)": 52.51,
1049
+ "step": 475,
1050
+ "train_speed(iter/s)": 0.108888
1051
+ },
1052
+ {
1053
+ "acc": 0.59179163,
1054
+ "epoch": 1.794392523364486,
1055
+ "grad_norm": 2.554819107055664,
1056
+ "learning_rate": 2.7730360865923956e-06,
1057
+ "loss": 1.47536173,
1058
+ "memory(GiB)": 52.51,
1059
+ "step": 480,
1060
+ "train_speed(iter/s)": 0.109034
1061
+ },
1062
+ {
1063
+ "acc": 0.58686528,
1064
+ "epoch": 1.8130841121495327,
1065
+ "grad_norm": 2.176454544067383,
1066
+ "learning_rate": 2.287057172336021e-06,
1067
+ "loss": 1.51853113,
1068
+ "memory(GiB)": 52.51,
1069
+ "step": 485,
1070
+ "train_speed(iter/s)": 0.10918
1071
+ },
1072
+ {
1073
+ "acc": 0.61157169,
1074
+ "epoch": 1.8317757009345794,
1075
+ "grad_norm": 2.2419204711914062,
1076
+ "learning_rate": 1.8468740210672076e-06,
1077
+ "loss": 1.45838099,
1078
+ "memory(GiB)": 52.51,
1079
+ "step": 490,
1080
+ "train_speed(iter/s)": 0.109326
1081
+ },
1082
+ {
1083
+ "acc": 0.604812,
1084
+ "epoch": 1.8504672897196262,
1085
+ "grad_norm": 2.1367015838623047,
1086
+ "learning_rate": 1.4529091286973995e-06,
1087
+ "loss": 1.42373133,
1088
+ "memory(GiB)": 52.51,
1089
+ "step": 495,
1090
+ "train_speed(iter/s)": 0.10947
1091
+ },
1092
+ {
1093
+ "acc": 0.59049854,
1094
+ "epoch": 1.8691588785046729,
1095
+ "grad_norm": 2.212156057357788,
1096
+ "learning_rate": 1.1055406300002347e-06,
1097
+ "loss": 1.47500782,
1098
+ "memory(GiB)": 52.51,
1099
+ "step": 500,
1100
+ "train_speed(iter/s)": 0.10961
1101
+ },
1102
+ {
1103
+ "epoch": 1.8691588785046729,
1104
+ "eval_acc": 0.5858449542660069,
1105
+ "eval_loss": 1.55006742477417,
1106
+ "eval_runtime": 60.5418,
1107
+ "eval_samples_per_second": 0.826,
1108
+ "eval_steps_per_second": 0.826,
1109
+ "step": 500
1110
+ },
1111
+ {
1112
+ "acc": 0.61764479,
1113
+ "epoch": 1.8878504672897196,
1114
+ "grad_norm": 2.375039577484131,
1115
+ "learning_rate": 8.0510193567086e-07,
1116
+ "loss": 1.4303463,
1117
+ "memory(GiB)": 52.51,
1118
+ "step": 505,
1119
+ "train_speed(iter/s)": 0.108303
1120
+ },
1121
+ {
1122
+ "acc": 0.60544062,
1123
+ "epoch": 1.9065420560747663,
1124
+ "grad_norm": 2.1975295543670654,
1125
+ "learning_rate": 5.518814123121885e-07,
1126
+ "loss": 1.48970194,
1127
+ "memory(GiB)": 52.51,
1128
+ "step": 510,
1129
+ "train_speed(iter/s)": 0.108451
1130
+ },
1131
+ {
1132
+ "acc": 0.60760684,
1133
+ "epoch": 1.925233644859813,
1134
+ "grad_norm": 2.173210859298706,
1135
+ "learning_rate": 3.4612210565528326e-07,
1136
+ "loss": 1.43905754,
1137
+ "memory(GiB)": 52.51,
1138
+ "step": 515,
1139
+ "train_speed(iter/s)": 0.108595
1140
+ },
1141
+ {
1142
+ "acc": 0.61625342,
1143
+ "epoch": 1.9439252336448598,
1144
+ "grad_norm": 2.543931245803833,
1145
+ "learning_rate": 1.8802150727962876e-07,
1146
+ "loss": 1.40175552,
1147
+ "memory(GiB)": 52.51,
1148
+ "step": 520,
1149
+ "train_speed(iter/s)": 0.108738
1150
+ },
1151
+ {
1152
+ "acc": 0.61394835,
1153
+ "epoch": 1.9626168224299065,
1154
+ "grad_norm": 2.0409328937530518,
1155
+ "learning_rate": 7.773136505700995e-08,
1156
+ "loss": 1.36281643,
1157
+ "memory(GiB)": 52.51,
1158
+ "step": 525,
1159
+ "train_speed(iter/s)": 0.108834
1160
+ },
1161
+ {
1162
+ "acc": 0.60506306,
1163
+ "epoch": 1.9813084112149533,
1164
+ "grad_norm": 2.187635898590088,
1165
+ "learning_rate": 1.5357537501159423e-08,
1166
+ "loss": 1.45838461,
1167
+ "memory(GiB)": 52.51,
1168
+ "step": 530,
1169
+ "train_speed(iter/s)": 0.10897
1170
+ },
1171
+ {
1172
+ "epoch": 1.9962616822429906,
1173
+ "eval_acc": 0.5853951117109012,
1174
+ "eval_loss": 1.550318956375122,
1175
+ "eval_runtime": 60.5393,
1176
+ "eval_samples_per_second": 0.826,
1177
+ "eval_steps_per_second": 0.826,
1178
+ "step": 534
1179
+ }
1180
+ ],
1181
+ "logging_steps": 5,
1182
+ "max_steps": 534,
1183
+ "num_input_tokens_seen": 0,
1184
+ "num_train_epochs": 2,
1185
+ "save_steps": 50,
1186
+ "stateful_callbacks": {
1187
+ "TrainerControl": {
1188
+ "args": {
1189
+ "should_epoch_stop": false,
1190
+ "should_evaluate": false,
1191
+ "should_log": false,
1192
+ "should_save": true,
1193
+ "should_training_stop": true
1194
+ },
1195
+ "attributes": {}
1196
+ }
1197
+ },
1198
+ "total_flos": 2.3344607126351155e+17,
1199
+ "train_batch_size": 1,
1200
+ "trial_name": null,
1201
+ "trial_params": null
1202
+ }
qwen2-vl-7b-instruct/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d6094c64b2cfee66d8dd4d404b6e3d87c884814eed3dac6a3df1297447208dc
3
+ size 7416