reuters-gpt2-text-gen

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

• Loss: 5.3505

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: 0.0005
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.8	2	5.1325
No log	2.0	5	4.5158
No log	2.8	7	4.3926
3.2046	4.0	10	4.1889
3.2046	4.8	12	4.1255
3.2046	6.0	15	4.0377
3.2046	6.8	17	4.0633
2.3854	8.0	20	4.0414
2.3854	8.8	22	4.0933
2.3854	10.0	25	4.0976
2.3854	10.8	27	4.1448
2.0076	12.0	30	4.2108
2.0076	12.8	32	4.2002
2.0076	14.0	35	4.2741
2.0076	14.8	37	4.2844
1.7505	16.0	40	4.3736
1.7505	16.8	42	4.3574
1.7505	18.0	45	4.4569
1.7505	18.8	47	4.4581
1.5509	20.0	50	4.4699
1.5509	20.8	52	4.5080
1.5509	22.0	55	4.5407
1.5509	22.8	57	4.6452
1.3673	24.0	60	4.5325
1.3673	24.8	62	4.6152
1.3673	26.0	65	4.7127
1.3673	26.8	67	4.6173
1.2093	28.0	70	4.6912
1.2093	28.8	72	4.7465
1.2093	30.0	75	4.7689
1.2093	30.8	77	4.7705
1.0458	32.0	80	4.8648
1.0458	32.8	82	4.9239
1.0458	34.0	85	4.9503
1.0458	34.8	87	4.8597
0.9114	36.0	90	4.9523
0.9114	36.8	92	4.9581
0.9114	38.0	95	5.0170
0.9114	38.8	97	4.9739
0.7897	40.0	100	4.9779
0.7897	40.8	102	4.9746
0.7897	42.0	105	5.1164
0.7897	42.8	107	5.0466
0.688	44.0	110	5.1557
0.688	44.8	112	5.1215
0.688	46.0	115	5.1176
0.688	46.8	117	5.1375
0.6066	48.0	120	5.1657
0.6066	48.8	122	5.1599
0.6066	50.0	125	5.1915
0.6066	50.8	127	5.1978
0.5272	52.0	130	5.2156
0.5272	52.8	132	5.2771
0.5272	54.0	135	5.2110
0.5272	54.8	137	5.2720
0.4696	56.0	140	5.2585
0.4696	56.8	142	5.2798
0.4696	58.0	145	5.2785
0.4696	58.8	147	5.2969
0.424	60.0	150	5.3045
0.424	60.8	152	5.3076
0.424	62.0	155	5.3178
0.424	62.8	157	5.3264
0.3941	64.0	160	5.3031
0.3941	64.8	162	5.3250
0.3941	66.0	165	5.3291
0.3941	66.8	167	5.3288
0.3715	68.0	170	5.3393
0.3715	68.8	172	5.3485
0.3715	70.0	175	5.3370
0.3715	70.8	177	5.3340
0.3608	72.0	180	5.3379
0.3608	72.8	182	5.3413
0.3608	74.0	185	5.3434
0.3608	74.8	187	5.3471
0.351	76.0	190	5.3487
0.351	76.8	192	5.3499
0.351	78.0	195	5.3504
0.351	78.8	197	5.3505
0.3516	80.0	200	5.3505

Framework versions

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