lemonhat/Qwen2.5-Coder-7B-Instruct-swe_30k_v2_tag5

swe_30k_v2_tag5

This model is a fine-tuned version of Qwen/Qwen2.5-Coder-7B-Instruct on the swe_30k_v2_tag5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4522

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: 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
0.6812	0.0524	100	0.5303
0.664	0.1049	200	0.5222
0.6851	0.1573	300	0.5144
0.6637	0.2098	400	0.5085
0.5823	0.2622	500	0.4992
0.6342	0.3146	600	0.4874
0.5819	0.3671	700	0.4845
0.5393	0.4195	800	0.4796
0.7043	0.4719	900	0.4728
0.4485	0.5244	1000	0.4708
0.606	0.5768	1100	0.4642
0.521	0.6293	1200	0.4612
0.542	0.6817	1300	0.4597
0.5452	0.7341	1400	0.4562
0.5425	0.7866	1500	0.4558
0.5805	0.8390	1600	0.4525
0.5275	0.8915	1700	0.4524
0.5267	0.9439	1800	0.4526
0.5343	0.9963	1900	0.4521

Framework versions

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