izambasiron/gemma4-2b-markdown-review
izambasiron/gemma4-2b-markdown-review is a 2 billion parameter Gemma 4 E2B model, fine-tuned by izambasiron for automated Markdown document review. This specialized model excels at identifying issues such as missing alt text, broken links, stale filenames, and formatting errors within Markdown content. With a 32768 token context length, it is optimized for precise, task-specific analysis of Markdown documents.
Loading preview...
Overview
This model, izambasiron/gemma4-2b-markdown-review, is a specialized 2 billion parameter Gemma 4 E2B model, fine-tuned using QLoRA (4-bit) for automated Markdown document review. It was trained on 168 curated examples and evaluated on 42 held-out examples over 3 epochs, achieving a final validation loss of 0.546.
Key Capabilities
- Automated Markdown Review: Specifically designed to identify common issues in Markdown files.
- Error Detection: Catches missing alt text, broken links, stale filenames, and various formatting problems.
- Task-Specific: Optimized for a narrow, focused task rather than general conversational AI.
Usage and Limitations
The model can be easily integrated using transformers for Python or deployed with Ollama / llama.cpp using provided GGUF files. It requires approximately 3 GB RAM for Q4_K_M GGUF or 10 GB for fp16 transformers. Due to its training on a limited dataset (168 examples), it may exhibit overfitting to specific review patterns. It is exclusively English-only and not suitable for general chat applications, being a highly task-specific fine-tune.