aczhu/qwen_reason_fc

sft

This model is a fine-tuned version of Qwen/Qwen2.5-14B-Instruct on the glaive_toolcall_100k and the bespoke_reasoning_17k datasets. It achieves the following results on the evaluation set:

• Loss: 0.3492

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 7
• gradient_accumulation_steps: 2
• total_train_batch_size: 14
• total_eval_batch_size: 7
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.4468	0.2307	500	0.3987
0.4457	0.4614	1000	0.3861
0.4197	0.6920	1500	0.3745
0.4264	0.9227	2000	0.3640
0.3188	1.1532	2500	0.3638
0.2938	1.3839	3000	0.3572
0.2891	1.6145	3500	0.3523
0.3013	1.8452	4000	0.3492

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0