Hugging Face's logo

dev2bit
/
es2bash-mt5

Text2Text Generation
Transformers
PyTorch
Spanish
mt5
code
bash
Model card Files Community
1
es2bash-mt5 / README.md
fbohorquez's picture
fbohorquez
Update README.md
fd1b1e8
preview code
|
raw
history blame
5.55 kB
metadata
license: apache-2.0
datasets:
  - dev2bit/es2bash
language:
  - es
pipeline_tag: text2text-generation
tags:
  - code
  - bash
widget:
  - text: Muestra el contenido de file.py que se encuentra en ~/project/
    example_title: cat
  - text: Lista los 3 primeros archivos en /bin
    example_title: ls
  - text: Por favor, cambia al directorio /home/user/project/
    example_title: cd
  - text: Lista todos los 谩tomos del universo
    example_title: noCommand
  - text: ls -lh
    example_title: literal
  - text: file.txt
    example_title: simple

es2bash-mt5: Modelo de traducci贸n de espa帽ol a Bashcd

Developed by dev2bit, es2bash-mt5 is a language transformer model that is capable of predicting the optimal Bash command given a natural language request in Spanish. This model represents a major advancement in human-computer interaction, providing a natural language interface for Unix operating system commands.

About the Model

es2bash-mt5 is a fine-tuning model based on mt5-small. It has been trained on the dev2bit/es2bash dataset, which specializes in translating natural language in Spanish into Bash commands.

This model is optimized for processing requests related to the commands:

  • cat
  • ls
  • cd

Usage

Below is an example of how to use es2bash-mt5 with the Hugging Face Transformers library:

from transformers import pipeline

translator = pipeline('translation', model='dev2bit/es2bash-mt5')

request = "listar los archivos en el directorio actual"
translated = translator(request, max_length=512)
print(translated[0]['translation_text'])

This will print the Bash command corresponding to the given Spanish request.

Contributions

We appreciate your contributions! You can help improve es2bash-mt5 in various ways, including:

  • Testing the model and reporting any issues or suggestions in the Issues section.
  • Improving the documentation.
  • Providing usage examples.

Desarrollado por dev2bit, es2bash-mt5 es un modelo transformador de lenguaje que tiene la capacidad de predecir el comando Bash 贸ptimo dada una solicitud en lenguaje natural en espa帽ol. Este modelo representa un gran avance en la interacci贸n humano-computadora, proporcionando una interfaz de lenguaje natural para los comandos del sistema operativo Unix.

Sobre el modelo

es2bash-mt5 es un modelo de ajuste fino basado en mt5-small. Ha sido entrenado en el conjunto de datos dev2bit/es2bash, especializado en la traducci贸n de lenguaje natural en espa帽ol a comandos Bash.

Este modelo est谩 optimizado para procesar solicitudes relacionadas con los comandos:

  • cat
  • ls
  • cd

Uso

A continuaci贸n, se muestra un ejemplo de c贸mo usar es2bash-mt5 con la biblioteca Hugging Face Transformers:

from transformers import pipeline

translator = pipeline('translation', model='dev2bit/es2bash-mt5')

request = "listar los archivos en el directorio actual"
translated = translator(request, max_length=512)
print(translated[0]['translation_text'])

Esto imprimir谩 el comando Bash correspondiente a la solicitud dada en espa帽ol.

Contribuciones

Agradecemos sus contribuciones! Puede ayudar a mejorar es2bash-mt5 de varias formas, incluyendo:

  • Probar el modelo y reportar cualquier problema o sugerencia en la secci贸n de Issues.
  • Mejorando la documentaci贸n.
  • Proporcionando ejemplos de uso.

This model is a fine-tuned version of google/mt5-small on the es2bash dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0928

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0001
  • train_batch_size: 8
  • eval_batch_size: 1
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 25

Training results

Training Loss Epoch Step Validation Loss
21.394 1.0 672 1.7470
2.5294 2.0 1344 0.6350
0.5873 3.0 2016 0.2996
0.3802 4.0 2688 0.2142
0.2951 5.0 3360 0.1806
0.225 6.0 4032 0.1565
0.2065 7.0 4704 0.1461
0.1944 8.0 5376 0.1343
0.174 9.0 6048 0.1281
0.1647 10.0 6720 0.1207
0.1566 11.0 7392 0.1140
0.1498 12.0 8064 0.1106
0.1382 13.0 8736 0.1076
0.1393 14.0 9408 0.1042
0.1351 15.0 10080 0.1019
0.13 16.0 10752 0.0998
0.1292 17.0 11424 0.0983
0.1265 18.0 12096 0.0973
0.1255 19.0 12768 0.0969
0.1216 20.0 13440 0.0956
0.1216 21.0 14112 0.0946
0.123 22.0 14784 0.0938
0.113 23.0 15456 0.0931
0.1185 24.0 16128 0.0929
0.1125 25.0 16800 0.0928

Framework versions

  • Transformers 4.29.2
  • Pytorch 2.0.1+cu117
  • Datasets 2.12.0
  • Tokenizers 0.13.3