Finance Lora Qwen2.5 1.5b

Finance Lora Qwen2.5 1.5b is a LoRA adapter designed for use with the Qwen2.5-1.5B-Instruct-4bit language model. It has been fine-tuned on the gbharti/finance-alpaca financial instruction dataset, with the training process resulting in a reduction of test perplexity from 6.303 (base) to 5.847 (tuned) across three seeds. The tool is intended for integration with MLX, and instructions are provided for downloading and using the adapter with the MLX library. Users can load the base model and apply the adapter through specified code snippets, enabling adaptation of the language model for financial instruction tasks. The model card does not specify the target audience, licensing terms, or pricing details. There is no information about additional features, integrations, or supported deployment platforms beyond its compatibility with MLX and the use of the Hugging Face Hub for distribution. The tool is positioned as a model adapter within the broader class of foundation models, specifically tailored for financial instruction data.