| license: apache-2.0 | |
| base_model: Qwen/Qwen2.5-0.5B-Instruct | |
| tags: | |
| - gguf | |
| - ffmpeg | |
| - command-generation | |
| - lm-studio | |
| - ollama | |
| # FFMPEG Command Generator (GGUF) | |
| Fine-tuned Qwen2.5-0.5B that generates FFMPEG commands from natural language with chain-of-thought reasoning. | |
| ## Quick Start | |
| ### LM Studio | |
| ```bash | |
| lms import kingjux/ffmpeg-command-generator-gguf | |
| ``` | |
| ### Ollama | |
| ```bash | |
| ollama run hf.co/kingjux/ffmpeg-command-generator-gguf | |
| ``` | |
| ## Example | |
| **Input:** "Convert video.mp4 to webm format" | |
| **Output:** | |
| ``` | |
| <think> | |
| Task: Convert MP4 to WebM | |
| - WebM uses VP9 video + Opus audio | |
| - Use -c:v libvpx-vp9 for video | |
| - Use -c:a libopus for audio | |
| </think> | |
| ffmpeg -i video.mp4 -c:v libvpx-vp9 -c:a libopus output.webm | |
| ``` | |
| ## Training | |
| - Base: Qwen2.5-0.5B-Instruct | |
| - Method: LoRA fine-tuning (r=16, alpha=32) | |
| - Dataset: 30 FFMPEG command examples with CoT reasoning | |
| - Trained on HuggingFace Jobs (T4 GPU) | |