rotating-head-lr-norm-gpt2-medium-wikitext

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.2154
• Accuracy: 0.4186
• Perplexity: 24.9126
• Bleu: 0.1314

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Perplexity	Bleu
5.9061	0.2806	500	5.7498	0.2230	314.1125	0.0496
4.8622	0.5612	1000	4.7414	0.2810	114.5910	0.0705
4.3006	0.8418	1500	4.2267	0.3182	68.4878	0.0834
3.9714	1.1223	2000	3.9429	0.3439	51.5654	0.0924
3.7835	1.4029	2500	3.7523	0.3629	42.6192	0.0969
3.6732	1.6835	3000	3.6293	0.3750	37.6861	0.1067
3.5764	1.9641	3500	3.5353	0.3848	34.3055	0.1124
3.4733	2.2447	4000	3.4822	0.3899	32.5321	0.1183
3.4163	2.5253	4500	3.4356	0.3946	31.0488	0.1253
3.3818	2.8058	5000	3.3806	0.4006	29.3886	0.1215
3.2827	3.0864	5500	3.3539	0.4028	28.6152	0.1308
3.2712	3.3670	6000	3.3233	0.4067	27.7517	0.1289
3.247	3.6476	6500	3.2908	0.4098	26.8652	0.1304
3.2203	3.9282	7000	3.2657	0.4126	26.1980	0.1278
3.1558	4.2088	7500	3.2440	0.4152	25.6357	0.1319
3.1152	4.4893	8000	3.2283	0.4169	25.2358	0.1301
3.1228	4.7699	8500	3.2154	0.4186	24.9126	0.1314

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0