Martinbvt/qwen25-1.5b-ifeval-fr
The Martinbvt/qwen25-1.5b-ifeval-fr model is a 1.5 billion parameter Qwen2.5-Instruct architecture, specifically fine-tuned for French language instruction following. It was trained using LoRA on 8,000 instructions verified by official IFEval checkers, making it particularly adept at understanding and executing French commands. This model is optimized for tasks requiring precise instruction adherence in French, leveraging its specialized training data.
Loading preview...
Model Overview
Martintbvt/qwen25-1.5b-ifeval-fr is a specialized language model based on the Qwen2.5-1.5B-Instruct architecture. It has been fine-tuned using the LoRA method on a dataset of 8,000 French instructions that were officially verified by IFEval checkers. This targeted training enhances its ability to accurately follow instructions in French.
Key Capabilities
- French Instruction Following: Excels at understanding and executing commands provided in French, due to its specific fine-tuning on verified French instructions.
- Qwen2 Architecture: Built upon the
Qwen2ForCausalLMarchitecture, ensuring robust language generation capabilities. - Efficient Fine-tuning: Utilizes LoRA (Low-Rank Adaptation) for efficient adaptation, making it suitable for deployment on resource-constrained environments like Mac M1.
- Context Length: Supports a context length of 32768 tokens.
Good For
- Applications requiring precise instruction adherence in French.
- Developing French-language chatbots or assistants that need to follow specific directives.
- Research and development focusing on instruction-tuned models for non-English languages, particularly French.
It's important to note that the le-leadboard/IFEval-fr test set was explicitly not used during the training process, ensuring unbiased evaluation against that benchmark.