gpt2_small_wiki_100M_32768_53

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.9882
• Accuracy: 0.3378

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: 32
• eval_batch_size: 32
• seed: 53
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 40000
• training_steps: 100000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
7.1917	0.22	2000	7.1253	0.1375
6.2932	0.43	4000	6.4682	0.1656
5.7724	0.65	6000	6.0204	0.1842
5.3495	0.87	8000	5.6941	0.1986
5.0008	1.08	10000	5.4465	0.2123
4.6744	1.3	12000	5.2044	0.2326
4.3954	1.52	14000	5.0246	0.2466
4.186	1.73	16000	4.8652	0.2584
4.0297	1.95	18000	4.7515	0.2660
3.8857	2.17	20000	4.6631	0.2734
3.81	2.38	22000	4.5863	0.2789
3.7289	2.6	24000	4.5271	0.2832
3.677	2.82	26000	4.4742	0.2878
3.6032	3.03	28000	4.4318	0.2922
3.5399	3.25	30000	4.3990	0.2955
3.5102	3.47	32000	4.3566	0.2981
3.484	3.68	34000	4.3380	0.3012
3.4487	3.9	36000	4.3095	0.3034
3.3679	4.12	38000	4.2899	0.3053
3.3619	4.33	40000	4.2583	0.3080
3.3495	4.55	42000	4.2440	0.3095
3.3216	4.77	44000	4.2131	0.3115
3.3056	4.98	46000	4.1926	0.3145
3.2263	5.2	48000	4.1758	0.3160
3.219	5.42	50000	4.1600	0.3178
3.2041	5.63	52000	4.1466	0.3192
3.1942	5.85	54000	4.1256	0.3210
3.1384	6.07	56000	4.1273	0.3214
3.1184	6.28	58000	4.1083	0.3233
3.1166	6.5	60000	4.0978	0.3241
3.1126	6.72	62000	4.0857	0.3253
3.106	6.93	64000	4.0709	0.3269
3.0349	7.15	66000	4.0753	0.3267
3.0382	7.36	68000	4.0661	0.3277
3.0407	7.58	70000	4.0545	0.3293
3.0384	7.8	72000	4.0445	0.3303
3.0227	8.01	74000	4.0465	0.3312
2.9699	8.23	76000	4.0399	0.3313
2.976	8.45	78000	4.0330	0.3322
2.9766	8.66	80000	4.0177	0.3337
2.9713	8.88	82000	4.0155	0.3338
2.9172	9.1	84000	4.0189	0.3341
2.917	9.31	86000	4.0145	0.3347
2.9207	9.53	88000	4.0109	0.3352
2.9169	9.75	90000	3.9986	0.3364
2.9102	9.96	92000	3.9945	0.3366
2.8663	10.18	94000	3.9954	0.3371
2.8715	10.4	96000	3.9949	0.3371
2.8695	10.61	98000	3.9913	0.3376
2.8658	10.83	100000	3.9882	0.3378

Framework versions

• Transformers 4.30.2
• Pytorch 2.9.0+cu128
• Datasets 4.1.1
• Tokenizers 0.13.3