repoGitLabTrust 82 · PrimaryPublished 1mo agoLive · 1mo ago
Erwan923/gpu-infra-lab
GPU Inference Reliability Lab — vLLM + Mistral-7B sur k3s/Scaleway L4, observabilité DCGM/Prometheus, SLO, runbooks, bench FP16/FP8 mesuré au watt.
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 49%We'll benchmark an Open weights LLM on any GPU you choose — drop your model + hardware and we'll run it. [D] →
- PossiblePossibly related (embedding) · 48%Kicking off GPU Mode [D] →
- PossiblePossibly related (embedding) · 46%If your GPU can run inference, it should be able to fine-tune too. [P] →
- PossiblePossibly related (embedding) · 45%Hy3 (295B MoE) and NVIDIA Nemotron-Labs-Audex-30B-A3B (audio-capable 30B MoE) GGUF quants →
- PossiblePossibly related (embedding) · 45%Measuring PCIe transfer under dual GPU with pipeline & tensor llama.cpp →
- PossiblePossibly related (embedding) · 50%AI-Assisted GPU Porting of a 250k Line Legacy Weather Simulation Code →
- PossiblePossibly related (embedding) · 56%Show HN: PantheonGPU – GPU health testing and AI workload benchmarking →
- PossiblePossibly related (embedding) · 45%Madlad builds homebrew GPU using 8,192 RISC-V chips →
Covers
Covers (incoming)
newsHy3 (295B MoE) and NVIDIA Nemotron-Labs-Audex-30B-A3B (audio-capable 30B MoE) GGUF quantsnewsMeasuring PCIe transfer under dual GPU with pipeline & tensor llama.cppnewsAI-Assisted GPU Porting of a 250k Line Legacy Weather Simulation CodenewsShow HN: PantheonGPU – GPU health testing and AI workload benchmarkingnewsMadlad builds homebrew GPU using 8,192 RISC-V chips
Related across the graph
newsShow HN: PantheonGPU – GPU health testing and AI workload benchmarkingnewsAI-Assisted GPU Porting of a 250k Line Legacy Weather Simulation CodenewsMadlad builds homebrew GPU using 8,192 RISC-V chipsnewsHy3 (295B MoE) and NVIDIA Nemotron-Labs-Audex-30B-A3B (audio-capable 30B MoE) GGUF quantsnewsKicking off GPU Mode [D]newsWe'll benchmark an Open weights LLM on any GPU you choose — drop your model + hardware and we'll run it. [D]newsMeasuring PCIe transfer under dual GPU with pipeline & tensor llama.cppnewsIf your GPU can run inference, it should be able to fine-tune too. [P]