A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 5, 18:30 UTC
[CI Failure] pin colmodernvbert revision (#38612)
Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io> Signed-off-by: wang.yuqi <noooop@126.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
wang.yuqi committed
Mar 31, 2026 at 18:54 UTC
719735d6c5f5dae14948c2150e6858351994b0b9
2 files changed
+9
tests/models/multimodal/pooling/test_colmodernvbert.py
+8
@@ -15,6 +15,10 @@ from vllm.entrypoints.pooling.scoring.utils import compute_maxsim_score
15
MODEL_NAME = "ModernVBERT/colmodernvbert-merged"
16
COLBERT_DIM = 128
17
DTYPE = "half"
18
+# Fixme:
19
+# Update colmodernvbert code to support the latest HF version
20
+# and remove revision set.
21
+REVISION = "4a0a9f3ac7a7992fec410bfa8e3d080ac9a5bcee"
22
23
24
# -----------------------------------------------------------------------
@@ -26,6 +30,7 @@ def test_colmodernvbert_text_token_embed(vllm_runner):
30
"""Text query produces per-token embeddings with shape (seq_len, 128)."""
31
with vllm_runner(
32
MODEL_NAME,
33
+ revision=REVISION,
34
runner="pooling",
35
dtype=DTYPE,
36
enforce_eager=True,
@@ -49,6 +54,7 @@ def test_colmodernvbert_text_relevance_ordering(vllm_runner):
54
55
with vllm_runner(
56
MODEL_NAME,
57
+ revision=REVISION,
58
runner="pooling",
59
dtype=DTYPE,
60
enforce_eager=True,
@@ -66,6 +72,7 @@ def test_colmodernvbert_text_late_interaction(vllm_runner):
72
73
with vllm_runner(
74
MODEL_NAME,
75
+ revision=REVISION,
76
runner="pooling",
77
dtype=DTYPE,
78
enforce_eager=True,
@@ -92,6 +99,7 @@ def test_colmodernvbert_image_token_embed(vllm_runner, image_assets):
99
"""Image input produces per-token embeddings including vision tokens."""
100
with vllm_runner(
101
MODEL_NAME,
102
+ revision=REVISION,
103
runner="pooling",
104
dtype=DTYPE,
105
enforce_eager=True,
tests/models/registry.py
+1
@@ -636,6 +636,7 @@ _LATE_INTERACTION_EXAMPLE_MODELS = {
636
# [Multimodal]
637
"ColModernVBertForRetrieval": _HfExamplesInfo(
638
"ModernVBERT/colmodernvbert-merged",
639
+ revision="4a0a9f3ac7a7992fec410bfa8e3d080ac9a5bcee",
640
),
641
"ColPaliForRetrieval": _HfExamplesInfo("vidore/colpali-v1.3-hf"),
642
"ColQwen3": _HfExamplesInfo(