thetmon/c20
The thetmon/c20 is a 4 billion parameter LoRA adapter fine-tuned from the Qwen/Qwen3-4B-Instruct-2507 base model. Developed by thetmon, this adapter is specifically optimized to enhance multi-turn agent task performance across household tasks (ALFWorld) and database operations (DBBench). It improves the model's ability to learn environment observation, action selection, tool use, and error recovery within complex multi-turn trajectories.
Loading preview...
Overview
The thetmon/c20 is a LoRA adapter (r=64) specifically fine-tuned from the Qwen/Qwen3-4B-Instruct-2507 base model. This adapter focuses on significantly improving the base model's capabilities in multi-turn agent tasks, particularly within the domains of household task execution (ALFWorld) and database operations (DBBench).
Key Capabilities
- Enhanced Multi-Turn Agent Performance: The adapter is trained to improve the model's ability to handle complex, sequential tasks requiring multiple interactions.
- Improved Environment Interaction: It enables better learning of environment observation and appropriate action selection.
- Advanced Tool Use: The fine-tuning process emphasizes effective tool utilization within agent trajectories.
- Error Recovery: The model is trained to recover from errors during multi-turn interactions, making it more robust for agentic workflows.
Training Details
This adapter was trained using LoRA (full precision base) with a maximum sequence length of 4096 over 3 epochs. The training objective applied loss to all assistant turns in the multi-turn trajectory, fostering comprehensive learning across the entire interaction sequence. The training data included u-10bei/sft_alfworld_trajectory_dataset_v5 and u-10bei/dbbench_sft_dataset_react_v4.
Good For
- Agentic Applications: Ideal for developers building AI agents that need to perform complex, multi-step tasks.
- Automated Task Execution: Suitable for scenarios requiring a model to interact with environments, select tools, and manage multi-turn dialogues to achieve specific goals.
- Research in Agent AI: Provides a fine-tuned component for exploring and developing more capable AI agents in household and database contexts.