Revert "[Bugfix] Restore name-based model detection for HF cache snapshot paths" (#6642)
This commit is contained in:
1 parent
7ba5e8a146
commit
7439ed41c2
2 files changed
+4
-66
No files matched your search
@@ -15,11 +15,7 @@ from transformers import PretrainedConfig, Qwen3OmniMoeConfig
|
||||
|
||||
from tests.helpers.stage_config import get_deploy_config_path, get_deploy_config_stage
|
||||
from vllm_omni.config import config_factory as config_factory_module
|
||||
from vllm_omni.config.config_factory import (
|
||||
StageConfigFactory,
|
||||
_materialize_object_storage_configs,
|
||||
_name_match_candidate,
|
||||
)
|
||||
from vllm_omni.config.config_factory import StageConfigFactory, _materialize_object_storage_configs
|
||||
from vllm_omni.config.endpoint_policy import EndpointRestriction, OmniServingCapability
|
||||
from vllm_omni.config.omni_config import VllmOmniConfig
|
||||
from vllm_omni.config.pipeline_registry import OMNI_PIPELINES, register_pipeline, resolve_pipeline_config
|
||||
@@ -2936,44 +2932,3 @@ class TestObjectStorageConfigResolution:
|
||||
|
||||
matching = "s3://any-bucket/my-cosyvoice3-model"
|
||||
assert StageConfigFactory.try_infer_model_type(model=matching, trust_remote_code=False) == "cosyvoice3"
|
||||
|
||||
|
||||
class TestNameMatchCandidateSnapshotPaths:
|
||||
"""Name-based matching must survive resolved HF cache snapshot paths."""
|
||||
|
||||
_SNAPSHOT_REV = "29e01c4e8d000f4bcd70751be16fa94bf3d85a18"
|
||||
|
||||
def test_candidate_recovers_repo_name_from_hf_cache_layout(self):
|
||||
path = f"/data/hub/models--FunAudioLLM--Fun-CosyVoice3-0.5B-2512/snapshots/{self._SNAPSHOT_REV}"
|
||||
assert _name_match_candidate(path) == "Fun-CosyVoice3-0.5B-2512"
|
||||
assert _name_match_candidate(path + "/") == "Fun-CosyVoice3-0.5B-2512"
|
||||
|
||||
def test_candidate_keeps_basename_for_non_cache_paths(self):
|
||||
assert _name_match_candidate("org/My-Model") == "My-Model"
|
||||
assert _name_match_candidate("/data/snapshots-of/My-Model") == "My-Model"
|
||||
# A ``snapshots`` parent without the ``models--<org>--<name>`` encoding
|
||||
# falls back to the basename, same as before.
|
||||
assert _name_match_candidate(f"/data/notcache/snapshots/{self._SNAPSHOT_REV}") == self._SNAPSHOT_REV
|
||||
|
||||
def test_candidate_recovers_bare_repo_name_from_hf_cache_layout(self):
|
||||
# Legacy un-namespaced repos (e.g. ``gpt2``) cache as ``models--<name>``.
|
||||
path = f"/data/hub/models--my-cosyvoice3-model/snapshots/{self._SNAPSHOT_REV}"
|
||||
assert _name_match_candidate(path) == "my-cosyvoice3-model"
|
||||
|
||||
def test_candidate_ignores_malformed_cache_repo_dirs(self):
|
||||
# Hub validation rejects ``--`` inside repo ids, so more than three
|
||||
# ``--``-separated segments cannot be a real cache dir; keep basename.
|
||||
assert _name_match_candidate("/hub/models--a--b--c/snapshots/abc123") == "abc123"
|
||||
assert _name_match_candidate("/hub/models--/snapshots/abc123") == "abc123"
|
||||
|
||||
def test_try_infer_model_type_matches_cosyvoice3_snapshot_dir(self, tmp_path):
|
||||
"""Regression for the HF-cache path shape: CosyVoice3 ships an empty
|
||||
config.json and relies on name matching, but a resolved snapshot dir
|
||||
basename is a bare revision hash, so basename-only matching (#5036)
|
||||
misclassified it as a diffusion pipeline and stage init crashed with
|
||||
"Diffusers pipeline index not found"."""
|
||||
snapshot = tmp_path / "hub" / "models--FunAudioLLM--Fun-CosyVoice3-0.5B-2512" / "snapshots" / self._SNAPSHOT_REV
|
||||
snapshot.mkdir(parents=True)
|
||||
(snapshot / "config.json").write_text("{}")
|
||||
|
||||
assert StageConfigFactory.try_infer_model_type(model=str(snapshot), trust_remote_code=False) == "cosyvoice3"
|
||||
@@ -69,31 +69,14 @@ def _materialize_object_storage_configs(model: str) -> str:
|
||||
|
||||
|
||||
def _name_match_candidate(model: str) -> str:
|
||||
"""Model-name component of a model reference, used for name-based matching.
|
||||
"""Last path component of a model reference, used for name-based matching.
|
||||
|
||||
Object-storage URIs and HF repo ids carry non-model segments (bucket name,
|
||||
organization) that must not participate in substring matching; e.g. a
|
||||
bucket named ``qwen3-tts-models`` holding a ``Qwen3-Omni`` checkpoint must
|
||||
not resolve to the ``qwen3_tts`` pipeline. Usually that means the last
|
||||
path component.
|
||||
|
||||
A resolved HF hub cache snapshot path is the exception: it ends in
|
||||
``models--<org>--<name>/snapshots/<revision>`` (or ``models--<name>`` for
|
||||
legacy un-namespaced repos such as ``gpt2``), so its basename is a bare
|
||||
revision hash with no model identity (models with an empty ``config.json``
|
||||
such as CosyVoice3 then lose their only detection route). Recover ``name``
|
||||
from the repo segment; the organization still stays out of the match. The
|
||||
two forms are exhaustive because hub validation rejects ``--`` inside repo
|
||||
ids; anything else keeps the plain basename.
|
||||
not resolve to the ``qwen3_tts`` pipeline.
|
||||
"""
|
||||
parts = [part for part in str(model).rstrip("/").split("/") if part]
|
||||
if not parts:
|
||||
return ""
|
||||
if len(parts) >= 3 and parts[-2] == "snapshots":
|
||||
repo_dir = parts[-3].split("--")
|
||||
if repo_dir[0] == "models" and len(repo_dir) in (2, 3) and all(repo_dir):
|
||||
return repo_dir[-1]
|
||||
return parts[-1]
|
||||
return model.rstrip("/").rsplit("/", 1)[-1]
|
||||
|
||||
|
||||
def with_trust_remote_code_override(
|
||||
|
||||
Reference in new issue
Block a user