RichardErkhov/jysssacc_-_opt-350m_fine_lr5e-05_bs4_epoch20_wd0.01-awq

Quantization made by Richard Erkhov.

Github

Discord

Request more models

opt-350m_fine_lr5e-05_bs4_epoch20_wd0.01 - AWQ

• Model creator: https://huggingface.co/jysssacc/
• Original model: https://huggingface.co/jysssacc/opt-350m_fine_lr5e-05_bs4_epoch20_wd0.01/

Original model description:

license: other base_model: facebook/opt-350m tags: - generated_from_trainer model-index: - name: opt-350m_fine_lr5e-05_bs4_epoch20_wd0.01 results: []

opt-350m_fine_lr5e-05_bs4_epoch20_wd0.01

This model is a fine-tuned version of facebook/opt-350m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 6.4402

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
3.6861	1.0	157	3.5351
2.8315	2.0	314	3.7727
1.9912	3.0	471	4.1666
1.4737	4.0	628	4.4532
1.1977	5.0	785	4.9565
0.8678	6.0	942	4.9429
0.8014	7.0	1099	5.3148
0.685	8.0	1256	5.4586
0.6793	9.0	1413	5.3924
0.6246	10.0	1570	5.6821
0.5972	11.0	1727	5.8316
0.5604	12.0	1884	5.8977
0.5404	13.0	2041	6.0070
0.5375	14.0	2198	5.9317
0.5035	15.0	2355	6.1104
0.5088	16.0	2512	6.2142
0.4827	17.0	2669	6.1858
0.4744	18.0	2826	6.3306
0.4641	19.0	2983	6.3937
0.4531	20.0	3140	6.4402

Framework versions

• Transformers 4.36.2
• Pytorch 2.0.1
• Datasets 2.16.1
• Tokenizers 0.15.0