gpt2_small_wiki_30M_32768_76

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

• Loss: 4.3903
• Accuracy: 0.3017

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: 76
• 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
7.2268	0.73	2000	7.1421	0.1366
6.3104	1.45	4000	6.4822	0.1646
5.7831	2.18	6000	6.0335	0.1827
5.3543	2.91	8000	5.7011	0.1977
5.0021	3.64	10000	5.4463	0.2118
4.6744	4.36	12000	5.2256	0.2291
4.3888	5.09	14000	5.0205	0.2467
4.1655	5.82	16000	4.8771	0.2570
3.9866	6.55	18000	4.7694	0.2649
3.85	7.27	20000	4.6884	0.2718
3.7685	8.0	22000	4.6215	0.2768
3.6677	8.73	24000	4.5771	0.2811
3.5625	9.45	26000	4.5399	0.2853
3.5002	10.18	28000	4.5030	0.2884
3.4703	10.91	30000	4.4786	0.2905
3.3872	11.64	32000	4.4574	0.2932
3.305	12.36	34000	4.4532	0.2940
3.307	13.09	36000	4.4341	0.2962
3.2702	13.82	38000	4.4149	0.2975
3.2003	14.55	40000	4.4322	0.2984
3.1349	15.27	42000	4.4093	0.3004
3.1545	16.0	44000	4.3903	0.3017
3.0843	16.73	46000	4.3940	0.3026
2.991	17.45	48000	4.4067	0.3030
2.9767	18.18	50000	4.4133	0.3040

Framework versions

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