MNLP_M3_mcqa_anti_overfit

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: 1.9834

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: 8
• eval_batch_size: 8
• seed: 42
• 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: linear
• lr_scheduler_warmup_ratio: 0.15
• num_epochs: 2
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss
3.7902	0.05	50	3.8886
3.8278	0.1	100	3.7602
3.52	0.15	150	3.3903
3.0491	0.2	200	2.9518
2.7477	0.25	250	2.7795
2.6077	0.3	300	2.6335
2.4566	0.35	350	2.4626
2.3592	0.4	400	2.3554
2.3056	0.45	450	2.2539
2.1646	0.5	500	2.1980
2.0673	0.55	550	2.1717
2.029	0.6	600	2.1452
2.0557	0.65	650	2.1068
2.0771	0.7	700	2.0875
2.0488	0.75	750	2.0660
2.0679	0.8	800	2.0526
2.063	0.85	850	2.0389
2.0567	0.9	900	2.0321
2.0714	0.95	950	2.0214
1.9788	1.0	1000	2.0160
1.9398	1.05	1050	2.0144
1.928	1.1	1100	2.0196
1.9747	1.15	1150	2.0117
2.0118	1.2	1200	1.9938
1.9771	1.25	1250	1.9881
1.9078	1.3	1300	1.9844
1.9312	1.35	1350	1.9838
2.0094	1.4	1400	1.9834

Framework versions

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