771fcdd5a980c35aeb433aaf5c9ba11d

This model is a fine-tuned version of distilbert/distilroberta-base on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 0.2493
• Data Size: 1.0
• Epoch Runtime: 7.6959
• Accuracy: 0.9646
• F1 Macro: 0.9694

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	1.7954	0	0.7974	0.2271	0.0617
No log	1	170	1.7494	0.0078	1.2264	0.2271	0.0617
No log	2	340	1.6950	0.0156	1.0034	0.2771	0.0723
No log	3	510	1.6024	0.0312	1.1935	0.2917	0.1775
No log	4	680	0.9255	0.0625	1.4415	0.7208	0.5931
0.0702	5	850	0.4376	0.125	1.9225	0.8792	0.7397
0.0702	6	1020	0.2226	0.25	2.7496	0.9437	0.9233
0.2657	7	1190	0.2360	0.5	4.4599	0.9437	0.9203
0.1723	8.0	1360	0.1594	1.0	7.9839	0.9688	0.9618
0.1121	9.0	1530	0.2252	1.0	7.8864	0.9542	0.9443
0.083	10.0	1700	0.2485	1.0	8.1102	0.9563	0.9397
0.036	11.0	1870	0.1957	1.0	8.0082	0.9688	0.9711
0.0288	12.0	2040	0.2493	1.0	7.6959	0.9646	0.9694

Framework versions

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