 Alexander Korolev
|
3d986923c7
fix: drop incompatible -q flag from container label query (docker -q + --format conflict)
|
преди 4 месеца |
|
Alexander Korolev
|
68087045df
fix: resolve_cluster_by_container_label now actually reads the sm121.cluster label from container metadata instead of inferring cluster from node ownership
|
преди 4 месеца |
|
Alexander Korolev
|
2946e07f79
fix: stop_cluster filters by cluster label to avoid cross-cluster interference
|
преди 4 месеца |
|
Alexander Korolev
|
d478170365
fix: switch MiMo recipe to mp backend, remove Ray-specific env vars
|
преди 4 месеца |
|
Alexander Korolev
|
5837205e68
feat: add MiMo-V2.5 Omni NVFP4 recipe (2× DGX Spark, Ray, 1M ctx)
|
преди 4 месеца |
|
Alexander Korolev
|
4936530710
feat: RecipeArgs enum — inline string or list for recipe args
|
преди 4 месеца |
|
Alexander Korolev
|
5931869d69
refactor: DistributedBackend as Option — remove None_ variant
|
преди 4 месеца |
|
Alexander Korolev
|
275e5f47ba
fix: keep serde(rename = "none") so existing recipes with 'distributed_backend = "none"' still parse
|
преди 4 месеца |
|
Alexander Korolev
|
09841103d8
fix: default DistributedBackend to None_ (absent means single-node)
|
преди 4 месеца |
|
Alexander Korolev
|
dbdbc7ed62
refactor: replace String distributed_backend with DistributedBackend enum
|
преди 4 месеца |
|
Alexander Korolev
|
6081ccf443
feat: mods system — entrypoint wrapping + global/recipe mods with volume mount
|
преди 4 месеца |
|
Alexander Korolev
|
09ac2dc3c4
fix(deepseek-v4-flash): revert wrong tp pp changes
|
преди 4 месеца |
|
Alexander Korolev
|
746049fc18
fix(cli): switch clap help texts to English
|
преди 4 месеца |
|
Alexander Korolev
|
aa34a575e4
fix(i18n): wire up unused Fluent keys in before_serve, host_keys, health, exec
|
преди 4 месеца |
|
Alexander Korolev
|
f55cd70082
refactor: replace #[allow] with structs for start_cluster and build_docker_command
|
преди 4 месеца |
|
Alexander Korolev
|
ce60c5c59f
feat: serve follows logs by default, returns Cluster for follow-up
|
преди 4 месеца |
|
Alexander Korolev
|
e84b676397
feat: serve follows logs by default, --no-follow to skip
|
преди 4 месеца |
|
Alexander Korolev
|
d6ccc9af9a
fix: use docker stop only, no rm step
|
преди 4 месеца |
|
Alexander Korolev
|
be65b44ddb
fix: stop containers in single rm -f step
|
преди 4 месеца |
|
Alexander Korolev
|
6ff1e710aa
feat: record cluster label on containers for auto-detection
|
преди 4 месеца |
|
Alexander Korolev
|
d05805c577
feat: add --cluster to docker stop and logs commands
|
преди 4 месеца |
|
Alexander Korolev
|
ef03b8c784
feat: auto-scale PP when pipeline_parallel_size == required_nodes
|
преди 4 месеца |
|
Alexander Korolev
|
04413a8991
feat: auto-scale TP when recipe runs on larger cluster
|
преди 4 месеца |
|
Alexander Korolev
|
25540b4f48
fix: allow recipe with fewer required_nodes to run on larger cluster
|
преди 4 месеца |
|
Alexander Korolev
|
b5615f604d
fix(recipe): switch GLM-5.2 from SGLang to vLLM
|
преди 4 месеца |
|
Alexander Korolev
|
3d35c82535
feat(recipe): add GLM-5.2 NVFP4 recipe (4-node SGLang)
|
преди 4 месеца |
|
Alexander Korolev
|
8f000e671a
fix(ssh): separate ssh_session_follow with no inactivity timeout
|
преди 4 месеца |
|
Alexander Korolev
|
9d54d3f94c
gemma4-12b: working recipe with distributed_backend=none
|
преди 4 месеца |
|
Alexander Korolev
|
03c293c4ca
refactor(docker): move health_check to submodule
|
преди 4 месеца |
|
Alexander Korolev
|
71855e9f45
feat(recipe): add Gemma 4 12B IT (MTP) recipe + --node flag for single-node serve
|
преди 4 месеца |
