try1

This model is a fine-tuned version of Anwaarma/edos_taskB_llama3b_merged2_FINAL on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0008
• Accuracy: 0.6330
• F1 Macro: 0.5964
• F1 Micro: 0.6330

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: 0.0002
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT 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: 0.06
• num_epochs: 20
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	F1 Micro
1.1785	1.8598	100	1.3641	0.5597	0.5106	0.5597
0.977	3.7103	200	1.2230	0.5905	0.5455	0.5905
0.833	5.5607	300	1.0872	0.6193	0.5723	0.6193
0.7542	7.4112	400	1.0395	0.6152	0.5523	0.6152
0.727	9.2617	500	0.9886	0.6502	0.5612	0.6502
0.7084	11.1121	600	0.9770	0.6523	0.5784	0.6523
0.7088	12.9720	700	0.9677	0.6502	0.5786	0.6502
0.7005	14.8224	800	0.9622	0.6523	0.5831	0.6523
0.6984	16.6729	900	0.9635	0.6543	0.5847	0.6543
0.6982	18.5234	1000	0.9632	0.6481	0.5721	0.6481

Framework versions

• PEFT 0.17.1
• Transformers 4.56.2
• Pytorch 2.8.0+cu126
• Datasets 4.1.1
• Tokenizers 0.22.0