Alexander Korolev
|
3aa9d380b2
feat(recipes): agent-serving profile + GMU/env fixes from benchmark findings
|
hai 2 meses |
Alexander Korolev
|
0362145ef5
docs(specs): investigation — production-3.8 image update failures
|
hai 2 meses |
Alexander Korolev
|
a8e4932fc4
feat(deepseek-v4): add eugr b12x recipe variant with baked draft-loader mod
|
hai 2 meses |
Alexander Korolev
|
5bc841f211
tmp: rename to simplify
|
hai 2 meses |
Alexander Korolev
|
3a783ee58e
Revert "feat(deepseek-v4): move :fixed overlay to production-3.8 base + enable DSPARK tiny decode"
|
hai 2 meses |
Alexander Korolev
|
72f40b3da2
feat(deepseek-v4): move :fixed overlay to production-3.8 base + enable DSPARK tiny decode
|
hai 2 meses |
Alexander Korolev
|
e6f70ecef1
fix(deepseek-v4): forbid echoing injected content to close TC-34
|
hai 2 meses |
Alexander Korolev
|
8269eeafca
fix(deepseek-v4): honor thinking_mode kwarg; harden tool template vs prompt injection
|
hai 2 meses |
Alexander Korolev
|
bd292d7508
fix(deepseek-v4): use canonical model-card dsv4 encoding over stale bundled copy
|
hai 2 meses |
Alexander Korolev
|
a83c61cd7e
fix(deepseek-v4): patch vendored dsv4 tokenizer reasoning_effort handling
|
hai 2 meses |
Alexander Korolev
|
1bae27bd44
fix(docker): complete DS4 src runtime lock with flashinfer transitive deps
|
hai 2 meses |
Alexander Korolev
|
5a6ec550d7
fix(ssh): pass through already single-quoted tokens in shq
|
hai 2 meses |
Alexander Korolev
|
0a21797aeb
refactor(docker): extract model path-reference resolver to path_ref submodule
|
hai 2 meses |
Alexander Korolev
|
3e508cc2c8
feat(recipe): DSpark k=5 fix + experimental from-source DS4 build
|
hai 2 meses |
Alexander Korolev
|
e300668348
feat(docker): authoritative build/sync source, single-node build clusters, chunked upload
|
hai 2 meses |
Alexander Korolev
|
2df0e1168f
chore: remove piolium audit output and gitignore it
|
hai 2 meses |
Alexander Korolev
|
df4149920b
fix: remediate F1 (image-sync host-key verification) and F2 (docker run quoting)
|
hai 2 meses |
Alexander Korolev
|
0181645b9e
fix: make IB config optional for single-node clusters
|
hai 2 meses |
Alexander Korolev
|
a3bd00eef8
fix: tmux serve quoting, venv setup, cluster probe on dgx3.lan
|
hai 2 meses |
Alexander Korolev
|
1b4d07309a
fix: make NCCL config optional for single-node, capture stderr, support --on for tmux
|
hai 2 meses |
Alexander Korolev
|
c2956d7e7f
fix(i18n): use proper Fluent plural selectors for tmux-health-all-stopped
|
hai 2 meses |
Alexander Korolev
|
63cc9a42f5
fix(health): only show running tmux sessions, summarize stopped
|
hai 2 meses |
Alexander Korolev
|
87d5b3e85e
feat(health): show tmux deployments in health command
|
hai 2 meses |
Alexander Korolev
|
c4291ee522
feat(recipe): upgrade laguna to vLLM 0.26.0 + flashinfer 0.6.16.dev20260729
|
hai 2 meses |
Alexander Korolev
|
d047c30925
chore: update state.yaml with final head
|
hai 2 meses |
Alexander Korolev
|
0816b7ad25
fix(recipe): change solar-open2 served-model-name to solar
|
hai 2 meses |
Alexander Korolev
|
afa6999213
feat(tmux): support inline venv_setup + venv_setup_script
|
hai 2 meses |
Alexander Korolev
|
5f646966b8
feat(tmux): add venv_setup field for declarative venv creation
|
hai 2 meses |
Alexander Korolev
|
d9b922e01b
feat(recipe): add venv field for declarative tmux venv activation
|
hai 2 meses |
Alexander Korolev
|
759b4b5579
feat(recipe): deploy laguna-s2.1 via tmux with host venv (Docker image has wrong flashinfer)
|
hai 2 meses |