[Model] Add Clef-Flash support (#61308)

Signed-off-by: Tianyu Guo <guoty@inferact.ai>
Signed-off-by: Isotr0py <Isotr0py@outlook.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Isotr0py <Isotr0py@outlook.com>
This commit is contained in:
Tianyu Guo
2026-10-13 05:46:37 +00:00
committed by GitHub
co-authored by Codex Isotr0py
parent 0d2689e0df
commit 380638afbb
11 changed files with 580 additions and 361 deletions
@@ -1,6 +1,6 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
"""The Clef-Omni pooler against a port of the checkpoint's joint schema head,
"""The shared Clef pooler against a port of the checkpoint's joint schema head,
which loops over questions where the pooler batches them."""
import math
@@ -11,7 +11,10 @@ import pytest
import torch
import torch.nn.functional as F
from vllm.model_executor.models.clef_omni import JointSchemaHead, JointSchemaPooler
from vllm.model_executor.models.clef_joint_schema import (
JointSchemaHead,
JointSchemaPooler,
)
from vllm.pooling_params import PoolingParams
from vllm.v1.pool.metadata import PoolingMetadata, PoolingStates
+4
View File
@@ -802,6 +802,10 @@ _MULTIMODAL_EXAMPLE_MODELS = {
"Salesforce/blip2-opt-2.7b",
extras={"6b": "Salesforce/blip2-opt-6.7b"},
),
"ClefFlashForJointSchema": _HfExamplesInfo(
"Cloudflare/clef-flash",
max_model_len=4096,
),
"Cohere2VisionForConditionalGeneration": _HfExamplesInfo(
"CohereLabs/command-a-vision-07-2025",
extras={"command-a-plus": "CohereLabs/command-a-plus-05-2026-bf16"},