LFM2-700M-grammar-correction : GGUF

How to use:

This model is fine-tuned on a grammar-correction dataset. Use the following prefix before the text you would like to correct as the user query:

Fix grammar only. Return the corrected sentence only: {text}

Also adjust the following parameters:

temperature = 0.3
min_p = 0.15
repetition_penalty = 1.05

This model was finetuned and converted to GGUF format using Unsloth.

Example usage with llama.cpp:

• For text only LLMs: ./llama.cpp/llama-cli -hf redromnon/LFM2-700M-grammar-correction --jinja
• For multimodal models: ./llama.cpp/llama-mtmd-cli -hf redromnon/LFM2-700M-grammar-correction --jinja

Available Model files:

• LFM2-700M.Q5_K_M.gguf
• LFM2-700M.Q8_0.gguf
• LFM2-700M.Q4_K_M.gguf

Note

The model's BOS token behavior was adjusted for GGUF compatibility. This was trained 2x faster with Unsloth