Note: More models will be added to this comparison.

See axolotl config

axolotl version: 0.8.0.dev0

mlflow_tracking_uri: http://127.0.0.1:7860
mlflow_experiment_name: Qwen2.5-QwQ-RP-Draft-v0.2-1.5B-LoRA

# Hugging Face saving config
hub_model_id: BeaverAI/Qwen2.5-QwQ-RP-Draft-v0.2-1.5B-LoRA-WS
hub_strategy: every_save

# Model checkpointing config
output_dir: ./Outputs/Qwen2.5-QwQ-RP-Draft-v0.2-1.5B-LoRA
resume_from_checkpoint:
save_steps: 10
save_safetensors: true
save_total_limit: 3
save_only_model: false

# Model architecture config
base_model: Qwen/Qwen2.5-1.5B-Instruct
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

# Mixed precision training config
bf16: true
fp16: false
tf32: false

# Model loading config
load_in_8bit: false
load_in_4bit: false
strict: false

# Sequence config
sequence_len: 8192
min_sample_len: 256
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
train_on_inputs: false
group_by_length: false

# LoRA adapter config
adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 128
lora_dropout: 0.125
peft_layers_to_transform:
peft_use_dora:
peft_use_rslora:
peft_layer_replication:
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj
lora_modules_to_save:

# Fix uninitialized tokens (such as <|start_header_id|> on the base L3 models)
fix_untrained_tokens:

# Dataset config
# https://github.com/xzuyn/axolotl/blob/prompt_formats/src/axolotl/prompt_strategies/customchatml-regex-last-only.py
datasets:
  - path: PJMixers-Dev/allura-org_gryphe-sonnet-3.5-charcards-names-added-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/anthracite-org_c2_logs_32k_llama3_qwen2_v1.3-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_aicg-logs-augmented-system-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_jannie-log-augmented-system-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_PIPPA-augmented-dedup-system-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/lemonilia_LimaRP-Only-NonSus-Simple-CustomShareGPT-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/MinervaAI_Aesir-Preview-Anon-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/NyxKrage_chub-logs-sharegpt-longest-CustomShareGPT-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/PocketDoc_Dans-Prosemaxx-Cowriter-XL-8192-shrunk-l3-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/PocketDoc_Dans-Personamaxx-Rainy-qwq-all-aphrodite-Shuffled
    split: train
    type: customchatml-regex-last-only
test_datasets:
  - path: PJMixers-Dev/allura-org_gryphe-sonnet-3.5-charcards-names-added-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/anthracite-org_c2_logs_32k_llama3_qwen2_v1.3-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_aicg-logs-augmented-system-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_jannie-log-augmented-system-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/grimulkan_PIPPA-augmented-dedup-system-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/lemonilia_LimaRP-Only-NonSus-Simple-CustomShareGPT-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/MinervaAI_Aesir-Preview-Anon-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/NyxKrage_chub-logs-sharegpt-longest-CustomShareGPT-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/PocketDoc_Dans-Prosemaxx-Cowriter-XL-8192-shrunk-l3-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
  - path: PJMixers-Dev/PocketDoc_Dans-Personamaxx-Rainy-qwq-all-aphrodite-Shuffled
    split: test
    type: customchatml-regex-last-only
val_set_size: 0
eval_strategy: steps
eval_steps: 10
dataset_prepared_path: ./00-Tokenized-Datasets/Qwen2.5-QwQ-Draft-v0.2-1.5B-customchatml-regex-last-only
shuffle_merged_datasets: true
dataset_processes:

# Training hyperparameters
num_epochs: 2
gradient_accumulation_steps: 1
micro_batch_size: 16  # x4 GPUs = 64
eval_batch_size: 16  # x4 GPUs = 64
warmup_steps: 0
optimizer: came_pytorch
optim_args:
optim_target_modules:
lr_scheduler: rex
learning_rate: 2e-5
cosine_min_lr_ratio:
loraplus_lr_ratio:
loraplus_lr_embedding:
weight_decay: 0.1
max_grad_norm: 1
logging_steps: 1

# Model optimization
gradient_checkpointing: unsloth
flash_attention: true
plugins:
  - axolotl.integrations.liger.LigerPlugin
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
cut_cross_entropy: true
liger_rope: true
liger_rms_norm: true
liger_layer_norm: true
liger_glu_activation: true
liger_cross_entropy: false
liger_fused_linear_cross_entropy: false
lora_mlp_kernel: false
lora_qkv_kernel: false
lora_o_kernel: false

# DeepSpeed
deepspeed: deepspeed_configs/zero3_bf16.json

# Garbage Collection
gc_steps: 1

# Debug config
debug: true
seed: 42

# Token config
special_tokens:
  eos_token: "<|endoftext|>"
  pad_token: "<|endoftext|>"
tokens:

Qwen2.5-QwQ-RP-Draft-v0.2-1.5B-LoRA-WS

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B-Instruct on the PJMixers-Dev/allura-org_gryphe-sonnet-3.5-charcards-names-added-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/anthracite-org_c2_logs_32k_llama3_qwen2_v1.3-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/grimulkan_aicg-logs-augmented-system-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/grimulkan_jannie-log-augmented-system-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/grimulkan_PIPPA-augmented-dedup-system-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/lemonilia_LimaRP-Only-NonSus-Simple-CustomShareGPT-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/MinervaAI_Aesir-Preview-Anon-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/NyxKrage_chub-logs-sharegpt-longest-CustomShareGPT-qwq-all-aphrodite-Shuffled, the PJMixers-Dev/PocketDoc_Dans-Prosemaxx-Cowriter-XL-8192-shrunk-l3-qwq-all-aphrodite-Shuffled and the PJMixers-Dev/PocketDoc_Dans-Personamaxx-Rainy-qwq-all-aphrodite-Shuffled datasets. It achieves the following results on the evaluation set:

• Loss: 1.7920

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 64
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_HF with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss
2.7253	0.0036	1	2.6463
2.4339	0.0362	10	2.3864
2.2821	0.0725	20	2.2571
2.207	0.1087	30	2.1771
2.1213	0.1449	40	2.1225
2.0926	0.1812	50	2.0870
2.0274	0.2174	60	2.0585
2.0292	0.2536	70	2.0338
2.0057	0.2899	80	2.0148
1.9841	0.3261	90	1.9975
1.9639	0.3623	100	1.9812
1.9303	0.3986	110	1.9688
1.9527	0.4348	120	1.9569
1.9389	0.4710	130	1.9459
1.9008	0.5072	140	1.9352
1.8996	0.5435	150	1.9259
1.8849	0.5797	160	1.9171
1.8793	0.6159	170	1.9105
1.8748	0.6522	180	1.9019
1.8697	0.6884	190	1.8959
1.8777	0.7246	200	1.8892
1.8731	0.7609	210	1.8833
1.8358	0.7971	220	1.8772
1.8483	0.8333	230	1.8729
1.8567	0.8696	240	1.8679
1.8461	0.9058	250	1.8632
1.8436	0.9420	260	1.8591
1.7992	0.9783	270	1.8545
1.7925	1.0145	280	1.8510
1.814	1.0507	290	1.8461
1.8249	1.0870	300	1.8431
1.7889	1.1232	310	1.8391
1.8116	1.1594	320	1.8367
1.7895	1.1957	330	1.8339
1.7951	1.2319	340	1.8300
1.7831	1.2681	350	1.8272
1.7919	1.3043	360	1.8250
1.7755	1.3406	370	1.8224
1.7895	1.3768	380	1.8209
1.7806	1.4130	390	1.8184
1.7808	1.4493	400	1.8156
1.7771	1.4855	410	1.8136
1.7636	1.5217	420	1.8118
1.7719	1.5580	430	1.8098
1.7716	1.5942	440	1.8074
1.7917	1.6304	450	1.8056
1.784	1.6667	460	1.8035
1.7846	1.7029	470	1.8019
1.7485	1.7391	480	1.8002
1.7381	1.7754	490	1.7987
1.764	1.8116	500	1.7972
1.744	1.8478	510	1.7964
1.7673	1.8841	520	1.7948
1.7926	1.9203	530	1.7935
1.7843	1.9565	540	1.7926
1.7689	1.9928	550	1.7920

Framework versions

• PEFT 0.14.0
• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.1