GitHub · Python
Blackwellboy/model-serving-minefield
Community registry of LLM serving-path traps that produce confidently wrong measurements: templates, tool parsers, reasoning fields, quant kernel paths, CUDA toolchains, KV allocation, eval harnesses, versioning. Symptom-first, with the check that catches each.
- Stars
- 177
- Forks
- 19
- Stars gained in 24 hours
- 0
- Forks gained in 24 hours
- 0
Topics: benchmarking, chat-template, cuda, debugging, llama-cpp, llm-serving, lm-eval, local-llm, model-evaluation, quantization
Latest data point: Nov 20, 2026 · 177 stars · 19 forks