richierich007/wingit-extractor-v1

TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jul 3, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The richierich007/wingit-extractor-v1 is an 8 billion parameter causal language model, fine-tuned from unsloth/llama-3.1-8b-instruct-bnb-4bit. It is specifically optimized for extracting high-value dialogue, concepts, and principles from dating and pickup transcriptions. This model excels at classifying content and generating structured JSONL training data for specialized applications.

Loading preview...

WingIt Extractor v1: Specialized Dialogue and Concept Extraction

The richierich007/wingit-extractor-v1 is an 8 billion parameter language model built upon the unsloth/llama-3.1-8b-instruct-bnb-4bit base. Its core purpose is to meticulously extract and categorize specific information from dating and pickup transcripts, focusing on high-value dialogue, underlying concepts, and principles.

Key Capabilities

  • Dialogue Extraction: Identifies and extracts conversational exchanges from raw transcript data.
  • Concept Identification: Pinpoints and categorizes behavioral concepts such as "frame control," "abundance," and "push-pull" within the text.
  • Content Classification: Accurately classifies segments of text as dialogue, concept, or principle.
  • Structured Data Generation: Capable of producing structured JSONL training data, making it suitable for further model development or analysis.

Training Methodology

The model was developed using a combination of Supervised Fine-Tuning (SFT) with 1,372 instruction examples and DPO (Direct Preference Optimization) to enhance the accuracy of its classification tasks.

Ideal Use Cases

This model is particularly well-suited for researchers, developers, or analysts working with specialized conversational data who need to automate the extraction and categorization of specific, high-value information from dating or pickup-related transcripts.