rotating-head-gp-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: 3.1985
• Accuracy: 0.4196
• Perplexity: 24.4954
• Bleu: 0.1339

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
5.9062	0.2806	500	5.7470	0.2234	313.2463	0.0493
4.8598	0.5612	1000	4.7428	0.2811	114.7554	0.0698
4.3025	0.8418	1500	4.2329	0.3170	68.9191	0.0834
3.9635	1.1223	2000	3.9291	0.3454	50.8590	0.0932
3.7769	1.4029	2500	3.7427	0.3636	42.2098	0.1020
3.6738	1.6835	3000	3.6225	0.3754	37.4295	0.1066
3.5744	1.9641	3500	3.5325	0.3845	34.2102	0.1118
3.456	2.2447	4000	3.4704	0.3902	32.1497	0.1139
3.3972	2.5253	4500	3.4190	0.3955	30.5384	0.1230
3.3654	2.8058	5000	3.3686	0.4007	29.0392	0.1230
3.247	3.0864	5500	3.3328	0.4043	28.0168	0.1247
3.2403	3.3670	6000	3.2985	0.4083	27.0714	0.1298
3.2167	3.6476	6500	3.2693	0.4112	26.2922	0.1288
3.1903	3.9282	7000	3.2456	0.4134	25.6768	0.1305
3.1212	4.2088	7500	3.2262	0.4161	25.1831	0.1325
3.0816	4.4893	8000	3.2128	0.4176	24.8480	0.1307
3.0917	4.7699	8500	3.1985	0.4196	24.4954	0.1339

Framework versions

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