Model Overview
This model, Qwen2.5-3B-Function-Calling-xLAM, is a 3.1 billion parameter language model developed by ermiaazarkhalili. It is a fine-tuned version of the Qwen/Qwen2.5-3B-Instruct base model, specifically optimized for function calling. The model was trained using Supervised Fine-Tuning (SFT) with LoRA (Low-Rank Adaptation) adapters on the Salesforce/xlam-function-calling-60k dataset.
Key Capabilities
- Function Calling Optimization: Specialized training on a dedicated function-calling dataset enhances its ability to interpret and generate function calls.
- Efficient Fine-Tuning: Utilizes LoRA with 4-bit NF4 quantization for efficient training and inference.
- Base Model Strength: Leverages the capabilities of the Qwen2.5-3B-Instruct architecture.
- Flexible Deployment: Available in multiple formats, including GGUF quantizations for CPU or mixed CPU/GPU inference.
Intended Use Cases
- Research: Ideal for studies on language model fine-tuning and function calling.
- Education: Suitable for learning and teaching purposes related to LLMs and SFT.
- Prototyping: Useful for developing and testing conversational AI agents that require function calling.
- Personal Projects: Can be integrated into various personal AI applications.
Limitations
- Primarily trained on English data.
- Knowledge cutoff is limited to the base model's training data.
- May exhibit hallucinations and is not extensively safety-tuned, requiring appropriate guardrails for production use.
- Fine-tuned with a context length limit of 2,048 tokens.