Tasmay-Tib/qwen2.5-1.5b-medical-sft-dare-resta
Tasmay-Tib/qwen2.5-1.5b-medical-sft-dare-resta is a 1.5 billion parameter language model, merged using the Task Arithmetic method with Qwen/Qwen2.5-1.5B-Instruct as its base. This model integrates specific medical and harmful content models, suggesting a focus on specialized domains. With a 32768 token context length, it is designed for applications requiring nuanced understanding within particular subject areas.
Loading preview...
Model Overview
This model, Tasmay-Tib/qwen2.5-1.5b-medical-sft-dare-resta, is a 1.5 billion parameter language model created by Tasmay-Tib. It was developed using the Task Arithmetic merge method, leveraging MergeKit to combine several pre-trained models.
Merge Details
The base model for this merge was Qwen/Qwen2.5-1.5B-Instruct. The merge process specifically incorporated:
outputs/part2/model_harmful_fullTasmay-Tib/qwen2.5-1.5b-medical-sft-dare-p03
Notably, the model_harmful_full component was included with a negative weight in the Task Arithmetic configuration, indicating an intent to potentially mitigate or adjust its influence. The model was configured with bfloat16 dtype.
Potential Use Cases
Given its specialized merge components, this model is likely intended for:
- Medical domain applications: The inclusion of
qwen2.5-1.5b-medical-sft-dare-p03suggests fine-tuning for medical text understanding or generation. - Content moderation research: The integration of a "harmful" model with a negative weight could be part of an exploration into content filtering or safety mechanisms.
- Specialized instruction following: Building upon the Qwen2.5-1.5B-Instruct base, it aims to provide instruction-tuned capabilities within its merged domains.