hiepnkv/qwen3-sft-scenario-to-code

scenario_to_selenium_Qwen3-14B_filtered

This model is a fine-tuned version of Qwen/Qwen3-14B on the sft_simle_forms_test_scenario_to_code_filter-executable dataset. It achieves the following results on the evaluation set:

• Loss: 0.0205

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: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss
0.0333	0.5006	51	0.0445
0.0189	1.0	102	0.0273
0.0196	1.5006	153	0.0240
0.0164	2.0	204	0.0227
0.0139	2.5006	255	0.0222
0.012	3.0	306	0.0213
0.0091	3.5006	357	0.0205
0.0103	4.0	408	0.0203
0.0113	4.5006	459	0.0204

Framework versions

• PEFT 0.15.1
• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1