ahammad115566/smeft-qwen-14b

Model Overview

The ahammad115566/smeft-qwen-14b is a specialized 14.8 billion parameter language model, fine-tuned from the Qwen2.5-14B-Instruct base model. Its primary domain is Standard Model Effective Field Theory (SMEFT), making it a unique tool for theoretical particle physics research.

Key Capabilities

• SMEFT Research Assistance: Designed to aid physicists in tasks such as understanding Beyond the Standard Model (BSM) phenomenology, manipulating operator bases, and analyzing Wilson coefficients.
• Domain-Specific Knowledge: Trained on SMEFT particle physics preprints, providing deep expertise in this complex field.
• Quantization for Efficiency: Utilizes 4-bit NF4 quantization (bitsandbytes) for inference, allowing for more efficient deployment.

Intended Use and Limitations

This model is intended for research assistance in SMEFT and related Effective Field Theory (EFT) frameworks. It aims to support physicists in their work, but it is crucial to note its limitations:

• Verification Required: The model may occasionally generate plausible-sounding but numerically incorrect values or operator identifications. All outputs must be verified against primary literature.
• Not a Substitute for Rigor: It is not designed to replace rigorous calculation or expert judgment.

Training Details

The model was fine-tuned using the LoRA method, with the LoRA adapters merged into the base model. The training data was sourced from SMEFT particle physics preprints, ensuring its specialized knowledge base.

License

This model is released under the Apache 2.0 License. Users should also be aware of the base model's (Qwen2.5-14B-Instruct) license terms.