Popular repositories Loading

1. hermes-concurrent-agents hermes-concurrent-agents Public

   Deploy concurrent Hermes Agent workers on unified-memory GPUs (GB10, DGX Spark) for maximum total tok/s. Profile-isolated, kanban-coordinated, crash-recovering.

   Python 72 8

2. llm-wiki_obsidian_hermes_r0b0tlabbra1n llm-wiki_obsidian_hermes_r0b0tlabbra1n Public

   Filesystem-first LLM-Wiki + Obsidian + Hermes Agent memory system. Markdown source of truth, SQLite FTS5 search, secret scanning, tier-based memory. Built for local LLM setups.

   Python 27 4

3. qwen36-35b-a3b-nvfp4-gb10-native-mtp qwen36-35b-a3b-nvfp4-gb10-native-mtp Public

   GB10 NVFP4 native MTP reproducibility pack for Qwen3.6-35B-A3B

   Python 13 1

4. minimax-m27-nvfp4-gb10-benchmark minimax-m27-nvfp4-gb10-benchmark Public

   MiniMax M2.7 NVFP4 dual-GB10 Blackwell benchmark: vLLM FlashInfer-CUTLASS, public data, HTML canvas report, and Docker runtime.

   HTML 13

5. deepseek-v4-flash-nvfp4-gb10-benchmark deepseek-v4-flash-nvfp4-gb10-benchmark Public

   DeepSeek-V4-Flash native Blackwell FP8 benchmark on dual DGX Spark GB10 (TP=2, MTP, RoCE). c=1 38.4 t/s, c=16 144.6 t/s aggregate.

   Python 10

6. nvidia-qwen-3.6-27B-sm121-nvfp4 nvidia-qwen-3.6-27B-sm121-nvfp4 Public

   NVIDIA Qwen3.6-27B NVFP4 on SM121 (GB10) — vLLM v0.24.0 with native NVFP4 KV cache via FlashInfer FA2 JIT. 67% more KV capacity than FP8.

   Python 7