rotating-head-lr-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: 3.2045
• Accuracy: 0.4189
• Perplexity: 24.6437
• Bleu: 0.1335

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
5.9064	0.2806	500	5.7488	0.2222	313.8089	0.0490
4.8623	0.5612	1000	4.7435	0.2805	114.8386	0.0709
4.3005	0.8418	1500	4.2307	0.3177	68.7669	0.0831
3.9583	1.1223	2000	3.9262	0.3464	50.7124	0.0916
3.7778	1.4029	2500	3.7522	0.3627	42.6128	0.1011
3.6805	1.6835	3000	3.6366	0.3738	37.9607	0.1052
3.5792	1.9641	3500	3.5434	0.3833	34.5829	0.1103
3.4655	2.2447	4000	3.4749	0.3900	32.2934	0.1168
3.4063	2.5253	4500	3.4235	0.3950	30.6766	0.1217
3.376	2.8058	5000	3.3767	0.4000	29.2745	0.1242
3.2591	3.0864	5500	3.3395	0.4039	28.2042	0.1258
3.2488	3.3670	6000	3.3072	0.4070	27.3098	0.1278
3.2244	3.6476	6500	3.2740	0.4109	26.4161	0.1309
3.1981	3.9282	7000	3.2526	0.4128	25.8571	0.1290
3.1294	4.2088	7500	3.2318	0.4156	25.3256	0.1293
3.0899	4.4893	8000	3.2180	0.4169	24.9787	0.1296
3.0993	4.7699	8500	3.2045	0.4189	24.6437	0.1335

Framework versions

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