17Lab/qwen14b-tool-routing-full-sft-s42
17Lab/qwen14b-tool-routing-full-sft-s42 is a 14.8 billion parameter language model fine-tuned from Qwen/Qwen2.5-14B. This model is specifically optimized for tool routing tasks, leveraging the assimilation_tool_routing_v1 dataset. It demonstrates high performance in its specialized domain, achieving a training loss of 0.0000 on the evaluation set. Its primary strength lies in accurately directing tasks to appropriate tools within an AI system.
Loading preview...
Model Overview
This model, 17Lab/qwen14b-tool-routing-full-sft-s42, is a specialized large language model built upon the Qwen/Qwen2.5-14B architecture. With 14.8 billion parameters and a context length of 32768 tokens, it has been meticulously fine-tuned for tool routing capabilities.
Key Capabilities
- Tool Routing Optimization: Specifically trained on the
assimilation_tool_routing_v1dataset to excel at directing user requests or tasks to the correct external tools or functions. - High Performance: Achieved a training loss of 0.0000 on the evaluation set, indicating strong specialization and accuracy in its intended function.
- Robust Base Model: Benefits from the advanced architecture and general language understanding of the Qwen2.5-14B foundation model.
Training Details
The model was trained with a learning rate of 2e-05, a total batch size of 64, and utilized a cosine learning rate scheduler over 3 epochs. The training process involved 225 steps, demonstrating rapid convergence to minimal loss values.
Good For
- Applications requiring precise and efficient tool selection or function calling based on natural language input.
- Integrating AI agents with external APIs, databases, or specialized modules where accurate routing is critical.
- Developing systems that automate complex workflows by intelligently dispatching tasks.