d3d167a908697d27a7a9855372a7c03e

This model is a fine-tuned version of facebook/opt-350m on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 0.3165
• Data Size: 1.0
• Epoch Runtime: 28.7909
• Accuracy: 0.9396
• F1 Macro: 0.9233

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	2.6821	0	1.6911	0.2104	0.1087
No log	1	170	3.4243	0.0078	1.8636	0.1437	0.0540
No log	2	340	1.1738	0.0156	2.3824	0.5917	0.4444
No log	3	510	0.6483	0.0312	3.2696	0.7729	0.6216
No log	4	680	0.5213	0.0625	4.6624	0.7625	0.7277
0.053	5	850	0.3146	0.125	6.3673	0.8896	0.7553
0.053	6	1020	0.3287	0.25	9.5504	0.9187	0.9097
0.2728	7	1190	0.3214	0.5	15.7646	0.9083	0.8775
0.2235	8.0	1360	0.2647	1.0	28.6645	0.9417	0.9006
0.1828	9.0	1530	0.3067	1.0	27.4585	0.9375	0.9370
0.0954	10.0	1700	0.4670	1.0	27.1525	0.9333	0.9223
0.1046	11.0	1870	0.3395	1.0	27.3568	0.95	0.9261
0.0591	12.0	2040	0.3165	1.0	28.7909	0.9396	0.9233

Framework versions

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