---
language:
- es
- en
license: apache-2.0
tags:
- code-generation
- text2bash
- sysadmin
- devops
- bash
- docker
- linux
base_model: Salesforce/codet5p-770m
datasets:
- GWHed/nl2bash
- MattCoddity/dockerNLcommands
- jrodriiguezg/MangoNLTBash
- b-mc2/cli-commands-explained
widget:
- text: >-
    Translate to Bash: busca todos los archivos pdf de más de 10MB y muévelos a
    /backup
  example_title: File Management
- text: >-
    Translate to Bash: reinicia los contenedores de docker que usen la imagen
    ubuntu
  example_title: Docker Logic
- text: 'Translate to Bash: deshaz el último commit pero mantén los cambios en local'
  example_title: Git Workflow
inference:
  parameters:
    max_length: 128
    num_beams: 2
---

# 🥭 MANGO-T5+ (770M) - Spanish Sysadmin Assistant

**Mango-T5+** is a fine-tuned CodeT5+ model specialized in translating natural language instructions into complex terminal commands (Bash, Docker, Git, Systemd, etc.). 

🚀 **Special Feature:** While based on a multilingual model, MANGO has been specifically optimized to understand **Spanish** instructions, including technical jargon and common sysadmin slang.

![Banner](https://img.shields.io/badge/Task-Text2Bash-orange)
![Precision](https://img.shields.io/badge/Precision-BF16-green)
![Hardware](https://img.shields.io/badge/Trained_on-NVIDIA_L4-blue)

## 💻 Usage

### Installation
```bash
pip install transformers torch sentencepiece safetensors

```

### Inference Code
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model_id = "jrodriiguezg/mango-t5-770m"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSeq2SeqLM.from_pretrained(model_id)

text = "Translate to Bash: bloquea la ip 192.168.1.50 en el firewall"
inputs = tokenizer(text, return_tensors="pt")

outputs = model.generate(**inputs, max_length=128)
command = tokenizer.decode(outputs[0], skip_special_tokens=True)

print(command)
# Output: sudo iptables -A INPUT -s 192.168.1.50 -j DROP
```
## Capabilities

Unlike generic LLMs, Mango-T5+ is focused on execution logic rather than chat. It excels at:

* **Complex Piping**: Chaining commands logically (e.g., find | xargs).
* **Docker Operations**: Advanced filtering and bulk actions.
* **Git Management**: distinguishing between soft/hard resets and branching.
* **System Administration**: systemd, journalctl, chmod, chown, user management.
  
### Example Outputs
| Input (Spanish)	| Generated Command |
|-------------------|-------------------|
|"Reinicia los contenedores que sean de ubuntu"	| `docker ps -q --filter ancestor=ubuntu |
|"Busca archivos modificados hoy" | find . -type f -mtime 0 |
|"Deshaz el último commit manteniendo cambios"	| git reset --soft HEAD~1 |

## Training Details
Dataset Composition

The model was trained on a mixed dataset of ~35,000 examples, combining:

* **NL2Bash**: The standard benchmark for bash translation.
* **Docker-NL**: Specific dataset for container orchestration.
* **TLDR Pages**: Inverse training using CLI tool descriptions.
* **Mango-DataSheet (Custom)**: A manually curated dataset in Spanish designed to fix common hallucinations, logic errors, and security pitfalls.

### Hyperparameters

* **Base Model**: Salesforce/codet5p-770m
* **Hardware**: NVIDIA L4 (24GB VRAM)
* **Precision**: BF16 (Brain Float 16)
* **Optimizer**: Adafactor
* **Batch Size**: 8 (Gradient Accumulation: 2)
* **Epochs**: 3

## Limitations & Safety

* **Context**: The model assumes a standard Linux environment (Fedora/Debian/Ubuntu).
* **Verification**: Always verify commands before execution, especially those involving file deletion (rm), disk formatting (mkfs, dd), or permission changes. The model generates the command, but the user is responsible for pulling the trigger.

## License

This model is fine-tuned from Salesforce/codet5p-770m and is distributed under the Apache 2.0 license.