gpt2_small_wiki_50M_32768_53

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

• Loss: 4.2036
• Accuracy: 0.3204

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
7.2067	0.43	2000	7.1418	0.1367
6.2966	0.87	4000	6.4781	0.1650
5.7649	1.3	6000	6.0221	0.1835
5.3429	1.73	8000	5.6953	0.1983
4.9856	2.17	10000	5.4411	0.2126
4.675	2.6	12000	5.2119	0.2312
4.3891	3.04	14000	5.0156	0.2474
4.1642	3.47	16000	4.8782	0.2576
4.0143	3.9	18000	4.7600	0.2660
3.8702	4.34	20000	4.6668	0.2731
3.785	4.77	22000	4.5986	0.2787
3.6662	5.2	24000	4.5459	0.2829
3.6283	5.64	26000	4.4943	0.2872
3.565	6.07	28000	4.4643	0.2901
3.5003	6.51	30000	4.4250	0.2936
3.4801	6.94	32000	4.3955	0.2965
3.3998	7.37	34000	4.3750	0.2982
3.3937	7.81	36000	4.3448	0.3012
3.2958	8.24	38000	4.3346	0.3031
3.3136	8.67	40000	4.3163	0.3045
3.2555	9.11	42000	4.3115	0.3061
3.2273	9.54	44000	4.2798	0.3087
3.2214	9.98	46000	4.2518	0.3107
3.1357	10.41	48000	4.2570	0.3122
3.1535	10.84	50000	4.2408	0.3134
3.0547	11.28	52000	4.2370	0.3144
3.0825	11.71	54000	4.2284	0.3155
3.0205	12.14	56000	4.2359	0.3159
3.0182	12.58	58000	4.2199	0.3167
3.0221	13.01	60000	4.2151	0.3185
2.9579	13.45	62000	4.2159	0.3185
2.9809	13.88	64000	4.2046	0.3193
2.8961	14.31	66000	4.2209	0.3192
2.9291	14.75	68000	4.2036	0.3204
2.8595	15.18	70000	4.2102	0.3205
2.8716	15.61	72000	4.2065	0.3206
2.8668	16.05	74000	4.2053	0.3216

Framework versions

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