Historial de Commits

Autor SHA1 Mensaje Fecha
  Alexander Korolev 08d1e9fe03 feat(serve): add --mirror — serve from a per-node local HF mirror hace 2 meses
  Alexander Korolev 4eff3e41b0 docs(skills): add nvidia-forum-reading skill hace 2 meses
  Alexander Korolev a432ba42be docs(skills): add sm121-launcher and vllm-mod-patches skills hace 2 meses
  Alexander Korolev 070bf6666a feat(deepseek-v4-eugr): optional reasoning_content key patch (default-off) hace 2 meses
  Alexander Korolev 0c7b29fbd6 feat(deepseek-v4-eugr): minimal reasoning_effort low/high/max patch (not full rewrite) hace 2 meses
  Alexander Korolev 25b18fcd56 fix(deepseek-v4): drafter b12x WO routing for real DSpark acceptance + GMU 0.78 hace 2 meses
  Alexander Korolev e296685e8c fix(deepseek-v4): let DSpark survive mixed-context batches (tool-eval-bench) hace 2 meses
  Alexander Korolev 5de97ee3aa fix(deepseek-v4): get DeepSeek-V4-Flash DSpark booting on production-3.8 (ds4) hace 2 meses
  Alexander Korolev d2026c2ddf feat(deepseek-v4): raise max-model-len to 1M + pin KV cache bytes (eugr) hace 2 meses
  Alexander Korolev 3aa9d380b2 feat(recipes): agent-serving profile + GMU/env fixes from benchmark findings hace 2 meses
  Alexander Korolev 0362145ef5 docs(specs): investigation — production-3.8 image update failures hace 2 meses
  Alexander Korolev a8e4932fc4 feat(deepseek-v4): add eugr b12x recipe variant with baked draft-loader mod hace 2 meses
  Alexander Korolev 5bc841f211 tmp: rename to simplify hace 2 meses
  Alexander Korolev 3a783ee58e Revert "feat(deepseek-v4): move :fixed overlay to production-3.8 base + enable DSPARK tiny decode" hace 2 meses
  Alexander Korolev 72f40b3da2 feat(deepseek-v4): move :fixed overlay to production-3.8 base + enable DSPARK tiny decode hace 2 meses
  Alexander Korolev e6f70ecef1 fix(deepseek-v4): forbid echoing injected content to close TC-34 hace 2 meses
  Alexander Korolev 8269eeafca fix(deepseek-v4): honor thinking_mode kwarg; harden tool template vs prompt injection hace 2 meses
  Alexander Korolev bd292d7508 fix(deepseek-v4): use canonical model-card dsv4 encoding over stale bundled copy hace 2 meses
  Alexander Korolev a83c61cd7e fix(deepseek-v4): patch vendored dsv4 tokenizer reasoning_effort handling hace 2 meses
  Alexander Korolev 1bae27bd44 fix(docker): complete DS4 src runtime lock with flashinfer transitive deps hace 2 meses
  Alexander Korolev 5a6ec550d7 fix(ssh): pass through already single-quoted tokens in shq hace 2 meses
  Alexander Korolev 0a21797aeb refactor(docker): extract model path-reference resolver to path_ref submodule hace 2 meses
  Alexander Korolev 3e508cc2c8 feat(recipe): DSpark k=5 fix + experimental from-source DS4 build hace 2 meses
  Alexander Korolev e300668348 feat(docker): authoritative build/sync source, single-node build clusters, chunked upload hace 2 meses
  Alexander Korolev 2df0e1168f chore: remove piolium audit output and gitignore it hace 2 meses
  Alexander Korolev df4149920b fix: remediate F1 (image-sync host-key verification) and F2 (docker run quoting) hace 2 meses
  Alexander Korolev 0181645b9e fix: make IB config optional for single-node clusters hace 2 meses
  Alexander Korolev a3bd00eef8 fix: tmux serve quoting, venv setup, cluster probe on dgx3.lan hace 2 meses
  Alexander Korolev 1b4d07309a fix: make NCCL config optional for single-node, capture stderr, support --on for tmux hace 2 meses
  Alexander Korolev c2956d7e7f fix(i18n): use proper Fluent plural selectors for tmux-health-all-stopped hace 2 meses