lemonhat/Qwen2.5-Coder-3B-Instruct-t2_25k_v2_tag5_processed

t2_25k_v2_tag5_processed

This model is a fine-tuned version of Qwen/Qwen2.5-Coder-3B-Instruct on the t2_25k_v2_tag5_processed dataset. It achieves the following results on the evaluation set:

• Loss: 0.2197

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: 5e-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 4
• total_eval_batch_size: 4
• 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
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.5549	0.0634	100	0.3669
0.3389	0.1268	200	0.2919
0.3116	0.1902	300	0.2736
0.3141	0.2536	400	0.2597
0.2972	0.3171	500	0.2492
0.3075	0.3805	600	0.2427
0.234	0.4439	700	0.2384
0.3061	0.5073	800	0.2332
0.3022	0.5707	900	0.2293
0.2999	0.6341	1000	0.2274
0.3069	0.6975	1100	0.2262
0.287	0.7609	1200	0.2220
0.2456	0.8244	1300	0.2204
0.2238	0.8878	1400	0.2200
0.2854	0.9512	1500	0.2198

Framework versions

• Transformers 4.46.1
• Pytorch 2.6.0+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3