MadeAgents/Hammer2.0-0.5b
Hammer2.0-0.5b by MadeAgents is a 0.5 billion parameter function-calling language model, fine-tuned on Qwen 2.5 series and Qwen 2.5 coder series. It is specifically optimized for robust function calling capabilities, trained using the APIGen Function Calling Datasets and xlam-irrelevance-7.5k. This model excels at enabling personalized, on-device agentic applications by accurately selecting and executing tools based on user requests, demonstrating strong performance across various function calling benchmarks.
Loading preview...
Hammer2.0-0.5b: Lightweight Function Calling for On-Device Agents
Hammer2.0-0.5b is a 0.5 billion parameter model developed by MadeAgents, specifically designed for robust function calling. It is part of a series of lightweight Hammer 2.0 models, including 1.5B, 3B, and 7B versions, all aimed at empowering developers to build personalized, on-device agentic applications.
Key Capabilities
- Exceptional Function Calling: Fine-tuned on the Qwen 2.5 series and Qwen 2.5 coder series, utilizing function masking techniques and trained with the APIGen Function Calling Datasets (60,000 samples) and xlam-irrelevance-7.5k.
- Strong Benchmark Performance: Achieves competitive performance on the Berkeley Function-Calling Leaderboard (BFCL-v3) and other academic benchmarks, demonstrating consistent and robust generalization ability.
- On-Device Agentic Applications: Optimized for scenarios requiring efficient tool selection and execution in resource-constrained environments.
Good For
- Developers building agentic applications that require precise function or tool invocation.
- Use cases demanding on-device processing where model size and efficiency are critical.
- Scenarios where robustness in function calling across diverse tasks is a priority.
- Integrating external tools and APIs into LLM-powered workflows.