metadata

license: mit
library_name: peft
tags:
  - trl
  - sft
  - generated_from_trainer
base_model: Aravindan/gpt2out
datasets:
  - generator
model-index:
  - name: output_dir
    results: []

output_dir

This model is a fine-tuned version of Aravindan/gpt2out on the generator dataset. It achieves the following results on the evaluation set:

Loss: 1.9619

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: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 10
total_train_batch_size: 80
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
training_steps: 1000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.6318	0.0147	30	2.4202
2.5147	0.0294	60	2.3425
2.4599	0.0440	90	2.2838
2.4009	0.0587	120	2.2386
2.394	0.0734	150	2.1971
2.3459	0.0881	180	2.1614
2.3057	0.1027	210	2.1324
2.3085	0.1174	240	2.1076
2.2675	0.1321	270	2.0891
2.2348	0.1468	300	2.0716
2.2167	0.1614	330	2.0594
2.1827	0.1761	360	2.0481
2.2049	0.1908	390	2.0390
2.1803	0.2055	420	2.0303
2.1709	0.2201	450	2.0250
2.1915	0.2348	480	2.0183
2.1583	0.2495	510	2.0120
2.168	0.2642	540	2.0072
2.1678	0.2788	570	2.0026
2.1545	0.2935	600	1.9988
2.1561	0.3082	630	1.9941
2.1442	0.3229	660	1.9913
2.1393	0.3375	690	1.9867
2.1489	0.3522	720	1.9834
2.1304	0.3669	750	1.9814
2.1175	0.3816	780	1.9783
2.113	0.3962	810	1.9753
2.1025	0.4109	840	1.9729
2.1181	0.4256	870	1.9711
2.0947	0.4403	900	1.9688
2.0868	0.4549	930	1.9665
2.1061	0.4696	960	1.9638
2.1096	0.4843	990	1.9619

Framework versions

PEFT 0.11.1
Transformers 4.41.2
Pytorch 2.1.2
Datasets 2.19.2
Tokenizers 0.19.1