chrissoria/catllm-json-formatter
The chrissoria/catllm-json-formatter is a 0.5 billion parameter Qwen2.5-0.5B-Instruct model fine-tuned by chrissoria to convert messy LLM classification outputs into a valid cat-llm JSON format. It specializes in parsing raw, potentially malformed text from other LLMs and restructuring it into a standardized JSON object. This model is designed to ensure 100% parse success for classification results, making it ideal for post-processing LLM outputs for structured data applications.
Loading preview...
CatLLM JSON Formatter Overview
This model, chrissoria/catllm-json-formatter, is a specialized 0.5 billion parameter Qwen2.5-0.5B-Instruct model developed by chrissoria. Its core function is to take raw, potentially malformed classification output from other Large Language Models and reliably convert it into a clean, standardized JSON format compatible with the cat-llm library.
Key Capabilities
- Robust JSON Formatting: Transforms unstructured or malformed LLM classification text into a consistent
{"1": "0", "2": "1", ...}JSON structure. - High Accuracy: Achieves 100% parse success and 98% exact match on a held-out test set, ensuring reliable output.
- Seamless Integration: Designed to be used automatically within the
cat-llmframework whenjson_formatter=Trueis enabled. - Efficient Size: Built on a 0.5B parameter base model, offering efficient performance for its specialized task.
Training Details
The model was fine-tuned using LoRA (r=16, alpha=32) on the Qwen/Qwen2.5-0.5B-Instruct base. Training involved 4,000 synthetic examples covering over 26 different messy output formats across 3 epochs.
Ideal Use Cases
This model is specifically engineered for scenarios where:
- You need to standardize and validate classification outputs from various LLMs.
- Ensuring parseable JSON from potentially inconsistent LLM responses is critical.
- Integrating LLM classification results into downstream applications that require structured data.