Python · MIT
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.
★ 173
Stars
17 forks · 21 open issues
+3.4/day
Growth · measured
Stars counted between two of our own snapshots roughly a week apart. The repository page gives the exact span.
48.6
Momentum / 100
see the breakdown below
3mo ago
Created
last push 6d ago
Star history
How this score was built
| Component | Weight | Earned | Why |
|---|---|---|---|
| Velocity | 55 | 12.4 | +3.4/day, log-scaled against a 750/day ceiling |
| Acceleration | 15 | 15.0 | last week ran 2.35× the monthly pace |
| Freshness | 15 | 14.7 | last push 6d ago; decays to zero at 180 days |
| Scale | 15 | 6.5 | 173 stars, log-scaled against 150k — stops a 3-star repo topping the board |
| Momentum | 100 | 48.6 | sum of the above |
Detail
| Topics | benchmarkingchat-templatecudadebuggingllama-cppllm-servinglm-evallocal-llmmodel-evaluationquantizationreasoning-modelsspeculative-decodingtroubleshootingvllm |
| Matched on | LLM & Inference (llm-serving) |
| Short window | +24 stars over 7 days = 3.4/day |
| Long window | +38 stars over 26 days = 1.5/day |
| Lifetime average | 1.71 stars/day over 101 days |
| First seen here | 2026-10-11 |
| Last read | 2d ago |