polynomial_1500_7e-4

This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.9716

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: 0.0007
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 320
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: polynomial
• lr_scheduler_warmup_steps: 250
• training_steps: 1500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
9.0482	0.2175	50	7.2874
6.6777	0.4350	100	6.1748
5.9153	0.6525	150	5.6097
5.4494	0.8699	200	5.2460
5.0771	1.0874	250	4.9104
4.7432	1.3049	300	4.5990
4.4572	1.5224	350	4.3180
4.2048	1.7399	400	4.0041
3.9347	1.9574	450	3.7463
3.6328	2.1749	500	3.5959
3.5206	2.3923	550	3.4919
3.424	2.6098	600	3.4093
3.35	2.8273	650	3.3413
3.2698	3.0448	700	3.2823
3.1187	3.2623	750	3.2438
3.0905	3.4798	800	3.2051
3.0669	3.6973	850	3.1715
3.0396	3.9147	900	3.1396
2.9253	4.1322	950	3.1144
2.8612	4.3497	1000	3.0928
2.8409	4.5672	1050	3.0711
2.8467	4.7847	1100	3.0497
2.826	5.0022	1150	3.0309
2.6716	5.2197	1200	3.0219
2.6832	5.4371	1250	3.0077
2.6746	5.6546	1300	2.9967
2.6572	5.8721	1350	2.9840
2.6081	6.0896	1400	2.9800
2.5443	6.3071	1450	2.9744
2.5481	6.5246	1500	2.9716

Framework versions

• Transformers 4.40.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1