Bump vllm to 0.6.6.post1

Signed-off-by: Jefferson Fialho <[email protected]>

Author: fialhocoelho

.github/workflows/tests.yaml

@@ -32,7 +32,7 @@ jobs:
         pyv: ["3.12"]
         vllm_version:
           # - "" # skip the pypi version as it will not work on CPU
-          - "git+https://github.com/vllm-project/[email protected].4"
+          - "git+https://github.com/vllm-project/[email protected].6.post1"
           - "git+https://github.com/vllm-project/vllm@main"
           - "git+https://github.com/opendatahub-io/vllm@main"
 

pyproject.toml

@@ -26,7 +26,7 @@ classifiers = [
 requires-python = ">=3.9"
 dynamic = ["version"]
 dependencies = [
-  "vllm>=0.6.4",
+  "vllm>=0.6.6.post1",
   "prometheus_client==0.21.0",
   "grpcio==1.67.0",
   "grpcio-health-checking==1.62.2",

src/vllm_tgis_adapter/grpc/grpc_server.py

@@ -62,12 +62,7 @@
     from grpc.aio import ServicerContext
     from vllm import CompletionOutput, RequestOutput
     from vllm.config import ModelConfig
-
-    try:
-        from vllm.engine.protocol import EngineClient
-    except ImportError:
-        # fallback for versions <=v0.6.1.post2
-        from vllm.engine.protocol import AsyncEngineClient as EngineClient
+    from vllm.engine.protocol import EngineClient
     from vllm.lora.request import LoRARequest
     from vllm.sequence import Logprob
     from vllm.transformers_utils.tokenizer import AnyTokenizer