RichardErkhov/Zagusan_-_Wikibot-3001-8bits

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Wikibot-3001 - bnb 8bits

• Model creator: https://huggingface.co/Zagusan/
• Original model: https://huggingface.co/Zagusan/Wikibot-3001/

Original model description:

license: mit base_model: Zagusan/Wikibot-3001 tags: - generated_from_trainer model-index: - name: Wikibot-3001 results: [] datasets: - mapama247/wikihow_es language: - es pipeline_tag: text-generation widget: - text: "Hola. ¿Cómo estás?" - text: "¿Cómo cocinar?" - text: "¿Cómo hornear un pastel?" - text: "¿Cómo hacer una investigación?" - text: "Método 1:"

Wikibot-3001

This model is a fine-tuned version of gpt2 on the mapama247/wikihow_es dataset. It achieves the following results on the evaluation set:

• Loss: 1.9422

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• distributed_type: tpu
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 12

Training results

Training Loss	Epoch	Step	Validation Loss
2.7432	1.0	5950	2.6109
2.3556	2.0	11900	2.3059
2.1872	3.0	17850	2.1738
2.0891	4.0	23800	2.0990
2.0306	5.0	29750	2.0579
1.9975	6.0	35700	2.0395
1.9909	7.0	5950	2.0208
1.8924	8.0	11900	1.9678
1.8281	9.0	17850	1.9422
1.8376	10.0	5950	1.9576
1.7802	11.0	11900	1.9212
1.7465	12.0	17850	1.8996

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3