d7a7897838a6251dac4e731978836789

This model is a fine-tuned version of meta-llama/Llama-3.2-3B on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 3.0863
• Data Size: 1.0
• Epoch Runtime: 49.5356
• Accuracy: 0.9208
• F1 Macro: 0.9047

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-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	10.0052	0	1.9847	0.15	0.0907
No log	1	170	19.5664	0.0078	2.4001	0.2667	0.1710
No log	2	340	9.2297	0.0156	7.8810	0.3875	0.2474
No log	3	510	3.5430	0.0312	13.8110	0.6937	0.6909
No log	4	680	1.8232	0.0625	16.7531	0.8896	0.7820
0.2706	5	850	1.8469	0.125	21.5650	0.8625	0.8695
0.2706	6	1020	1.5235	0.25	28.0599	0.9167	0.9238
1.4349	7	1190	2.2822	0.5	41.7559	0.8583	0.8786
1.1092	8.0	1360	1.1208	1.0	66.2516	0.9375	0.9461
0.8568	9.0	1530	1.0646	1.0	47.8734	0.9458	0.9497
0.7157	10.0	1700	1.7308	1.0	47.9386	0.95	0.9490
0.48	11.0	1870	1.2764	1.0	47.4131	0.9563	0.9524
0.3512	12.0	2040	2.1171	1.0	51.9403	0.9375	0.9368
0.3043	13.0	2210	3.0863	1.0	49.5356	0.9208	0.9047

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1