biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Core] Remove tokenizer group in vLLM (#24078)

Browse Source 
Signed-off-by: Zhuohan Li <zhuohan123@gmail.com>
This commit is contained in:
Zhuohan Li
2025-09-17 01:42:59 -07:00
committed by GitHub
parent c15309a730
commit 6c47f6bfa4
49 changed files with 276 additions and 934 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
tests/entrypoints/openai/test_lora_adapters.py
View File

@@ -67,12 +67,6 @@ def server_with_lora_modules_json(request, monkeypatch_module,
"base_model_name": MODEL_NAME
}
lora_module_2 = {
"name": "zephyr-lora2",
"path": zephyr_lora_files,
"base_model_name": MODEL_NAME
}
args = [
# use half precision for speed and memory savings in CI environment
"--dtype",
@@ -84,7 +78,6 @@ def server_with_lora_modules_json(request, monkeypatch_module,
"--enable-lora",
"--lora-modules",
json.dumps(lora_module_1),
json.dumps(lora_module_2),
"--max-lora-rank",
"64",
"--max-cpu-loras",
@@ -121,7 +114,6 @@ async def test_static_lora_lineage(client: openai.AsyncOpenAI,
for lora_model in lora_models)
assert all(lora_model.parent == MODEL_NAME for lora_model in lora_models)
assert lora_models[0].id == "zephyr-lora"
assert lora_models[1].id == "zephyr-lora2"
@pytest.mark.asyncio
@@ -209,7 +201,7 @@ async def test_dynamic_lora_badrequests(client: openai.AsyncOpenAI, tmp_path,
@pytest.mark.asyncio
async def test_multiple_lora_adapters(client: openai.AsyncOpenAI, tmp_path,
zephyr_lora_files):
"""Validate that many loras can be dynamically registered and inferenced
"""Validate that many loras can be dynamically registered and inferenced
with concurrently"""
# This test file configures the server with --max-cpu-loras=2 and this test