#171 · Primary category: MLOps & Evaluation
OpenPipe
Turn expensive prompts into cheap fine-tuned models
Project last updated:05/25/24
GitHub Stars
2.8K
Forks
178
Contributors
13
License
Apache-2.0
Why we included this project
Teams that send thousands of similar requests to large frontier models often end up paying per-token rates for repeated work. OpenPipe gives you a different path: log your real traffic through its OpenAI-compatible endpoint, then use that logged data to fine-tune smaller, cheaper models like Mistral and Llama 2, and swap between them by changing a single model name. It also includes an evaluation step that compares a fine-tuned model's output against the base model on your own logs, so you can check quality before you commit. One honest caveat for self-hosters: the maintainers paused work on this open-source release to integrate proprietary code, so treat the current snapshot as a proven reference architecture rather than an actively maintained codebase.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
unsloth
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, FLUX and more.
LlamaFactory
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
airflow
Apache Airflow - A platform to programmatically author, schedule, and monitor workflows
langfuse
🪢 Open source AI engineering platform: LLM evals, observability, metrics, prompt management, playground, datasets. Integrates with OpenTelemetry, LangChain, OpenAI SDK, LiteLLM, and more. 🍊YC W23
netron
Visualizer for neural network, deep learning and machine learning models