adeljebali/llama3.2-3B-instruct

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:May 1, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

The adeljebali/llama3.2-3B-instruct is a 3.2 billion parameter instruction-tuned language model developed by adeljebali, finetuned from unsloth/Llama-3.2-3B-Instruct-bnb-4bit. This model is specifically designed and optimized for Grammatical Error Correction (GEC) in French. It leverages Unsloth for faster training and offers a 32768 token context length, making it suitable for French language processing tasks requiring high accuracy in grammar and style.

Loading preview...

Model Overview

The adeljebali/llama3.2-3B-instruct is a 3.2 billion parameter instruction-tuned language model developed by adeljebali. It is finetuned from the unsloth/Llama-3.2-3B-Instruct-bnb-4bit base model, utilizing Unsloth and Huggingface's TRL library for accelerated training.

Key Capabilities

  • French Grammatical Error Correction (GEC): The primary specialization of this model is to identify and correct grammatical errors within French text.
  • Llama 3.2 Architecture: Built upon the Llama 3.2 series, inheriting its foundational language understanding capabilities.
  • Optimized Training: Training was conducted using Unsloth, which facilitates faster fine-tuning processes.
  • Context Length: Supports a context window of 32768 tokens, allowing for processing of longer French texts for GEC tasks.

Use Cases

This model is particularly well-suited for applications requiring robust French language processing, specifically:

  • Automated French Proofreading: Correcting grammar, spelling, and style in French documents.
  • Content Creation Assistance: Helping writers improve the grammatical quality of their French content.
  • Educational Tools: Providing feedback on French writing for learners.

Licensing

The model is released under the Apache-2.0 license.