LequeuISIR/AU-clarification_gemma-2-9b-it
LequeuISIR/AU-clarification_gemma-2-9b-it is a 9 billion parameter Gemma-2 model, fine-tuned by LequeuISIR on the GDN-CC dataset. This model is specifically optimized for Argumentative Unit Clarification in French, designed to correct grammar, syntax, and add context to opinion segments. It is identified as the best model for this specific task, used to annotate the GDN-CC-large corpus.
Loading preview...
Model Overview
LequeuISIR/AU-clarification_gemma-2-9b-it is a 9 billion parameter Gemma-2 model, specifically fine-tuned for Argumentative Unit Clarification in French. Developed by LequeuISIR, this model leverages the GDN-CC dataset, a manually-curated collection of contributions from the French Grand Débat National, to achieve its specialized performance. It is recognized as the leading model for this task and was instrumental in annotating the GDN-CC-large corpus.
Key Capabilities
- Argumentative Unit Clarification: Rewrites segments of opinion text to enhance clarity.
- Grammar and Syntax Correction: Automatically corrects spelling, grammar, and syntax errors within text segments.
- Contextual Integration: Incorporates relevant context from the initial text to ensure standalone comprehensibility of the clarified segment.
- French Language Optimization: Specifically trained and optimized for processing French opinion texts.
Use Cases
This model is ideal for preprocessing noisy, multi-topic contributions from large-scale citizen consultations or public forums. Its primary application is to transform these into structured, self-contained argumentative units, making them suitable for downstream analysis such as topic modeling and political analysis. The model is designed to be run efficiently, with recommendations for use with the vLLM framework.