GPT2-APPS

This model is a fine-tuned version of openai-community/gpt2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9545

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss
1.4339	0.04	200	1.2470
1.0147	0.08	400	1.1798
1.0875	0.12	600	1.1233
1.1918	0.16	800	1.0979
0.9112	0.2	1000	1.0793
0.9272	0.24	1200	1.0558
0.9595	0.28	1400	1.0268
0.9456	0.32	1600	1.0209
0.8592	0.36	1800	1.0090
0.8609	0.4	2000	1.0105
0.9569	0.44	2200	0.9996
0.9371	0.48	2400	0.9909
0.938	0.52	2600	0.9757
0.7827	0.56	2800	0.9976
0.8866	0.6	3000	0.9727
1.0179	0.64	3200	0.9729
0.7923	0.68	3400	0.9698
0.8139	0.72	3600	0.9635
0.8627	0.76	3800	0.9573
0.866	0.8	4000	0.9618
0.7777	0.84	4200	0.9557
0.8119	0.88	4400	0.9571
0.8964	0.92	4600	0.9548
0.9	0.96	4800	0.9546
0.8928	1.0	5000	0.9545

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2