library_name: transformers
license: apache-2.0
datasets:
- ThinkAgents/Function-Calling-with-Chain-of-Thoughts
base_model:
- meta-llama/Llama-3.2-1B-Instruct
Model Card for Model ID
Model Details
Model Description
ThinkAgent-1B is a fine-tuned version of meta-llama/Llama-3.2-1B-Instruct, specifically optimized for function calling tasks. The model is trained to improve Abstract Syntax Tree (AST) accuracy, ensuring more precise function call generations.
This model was developed as a graduation project by:
from the University of Khartoum, Electrical and Electronics Engineering Department.
For future work, further alignment of the model could be explored to enhance its function-calling capabilities.
Model Details
- Developed by: Ayman Tarig, Tasneem Midhat
- Model type: Fine-tuned language model for function calling
- Language(s) (NLP): English
- License: Apache-2.0
- Finetuned from model: meta-llama/Llama-3.2-1B-Instruct
Uses
Direct Use
The model can be used as-is for function-calling tasks, providing structured outputs for function calls based on natural language prompts.
Downstream Use
The model can be further fine-tuned for more specific function-calling applications, such as integrating into APIs or smart assistants.
Out-of-Scope Use
The model is not intended for general-purpose conversation, creative writing, or any tasks beyond function calling and structured response generation.
Bias, Risks, and Limitations
- The model is fine-tuned for function calling and may not generalize well outside this domain.
- The model’s outputs depend on the provided function definitions; incorrect function descriptions may lead to incorrect outputs.
- Potential biases inherited from the base model and dataset remain.
Recommendations
Users should validate the function call outputs before execution to avoid errors or unintended consequences.
Training Details
Training Data
The model was fine-tuned using the ThinkAgents/Function-Calling-with-Chain-of-Thoughts dataset. The dataset includes reasoning chains distilled from deepseek-ai/DeepSeek-R1-Distill-Llama-8B to enhance function-calling accuracy.