MNLP_M3_mcqa_sft_model

This model is a fine-tuned version of AnnaelleMyriam/SFT_M3_model on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5993

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.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: 0.05
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.3535	0.1352	250	0.4926
0.4864	0.2703	500	0.3696
0.342	0.4055	750	0.3518
0.3763	0.5407	1000	0.3259
0.3566	0.6759	1250	0.3335
0.2901	0.8110	1500	0.3195
0.3235	0.9462	1750	0.3060
0.2315	1.0811	2000	0.3930
0.2842	1.2163	2250	0.3920
0.2183	1.3514	2500	0.3796
0.1824	1.4866	2750	0.3979
0.1877	1.6218	3000	0.4335
0.1821	1.7570	3250	0.3981
0.2364	1.8921	3500	0.3922
0.1339	2.0270	3750	0.4119
0.1073	2.1622	4000	0.5467
0.0722	2.2974	4250	0.5596
0.113	2.4325	4500	0.5158
0.1467	2.5677	4750	0.4852
0.1675	2.7029	5000	0.5103
0.101	2.8381	5250	0.5661
0.1935	2.9732	5500	0.4946
0.1069	3.1081	5750	0.5844
0.0799	3.2433	6000	0.5681
0.0803	3.3785	6250	0.5795
0.0744	3.5137	6500	0.5935
0.0464	3.6488	6750	0.6010
0.0643	3.7840	7000	0.6009
0.0871	3.9192	7250	0.5993

Framework versions

• PEFT 0.15.2
• Transformers 4.52.4
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1