b7a582f21c83ad321f36f9a283975348

This model is a fine-tuned version of openai-community/gpt2-medium on the nyu-mll/glue [mnli] dataset. It achieves the following results on the evaluation set:

• Loss: 0.7697
• Data Size: 1.0
• Epoch Runtime: 1638.4196
• Accuracy: 0.8150
• F1 Macro: 0.8144
• Rouge1: 0.8147
• Rouge2: 0.0
• Rougel: 0.8146
• Rougelsum: 0.8150

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	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	14.8409	0	14.9453	0.3196	0.1808	0.3198	0.0	0.3196	0.3198
1.8286	1	12271	0.8433	0.0078	27.2457	0.6223	0.6162	0.6224	0.0	0.6222	0.6224
0.7745	2	24542	0.7041	0.0156	40.7216	0.7042	0.6999	0.7039	0.0	0.7041	0.7044
0.6389	3	36813	0.5944	0.0312	65.8869	0.7579	0.7541	0.7579	0.0	0.7580	0.7581
0.5844	4	49084	0.5385	0.0625	117.0396	0.7857	0.7859	0.7857	0.0	0.7858	0.7856
0.5095	5	61355	0.5024	0.125	215.3690	0.8031	0.8023	0.8031	0.0	0.8032	0.8031
0.4912	6	73626	0.4705	0.25	407.5710	0.8162	0.8152	0.8163	0.0	0.8162	0.8162
0.4033	7	85897	0.4502	0.5	799.9647	0.8255	0.8239	0.8255	0.0	0.8255	0.8258
0.3615	8.0	98168	0.4444	1.0	1587.5176	0.8310	0.8315	0.8310	0.0	0.8308	0.8310
0.2458	9.0	110439	0.5148	1.0	1591.8310	0.8250	0.8247	0.8248	0.0	0.8251	0.8248
0.2038	10.0	122710	0.5748	1.0	1621.9843	0.8226	0.8217	0.8226	0.0	0.8227	0.8226
0.1534	11.0	134981	0.6736	1.0	1640.1331	0.8271	0.8260	0.8269	0.0	0.8270	0.8271
0.1075	12.0	147252	0.7697	1.0	1638.4196	0.8150	0.8144	0.8147	0.0	0.8146	0.8150

Framework versions

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