[Deprecation][2/N] Replace --task with --runner and --convert (#21470)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Cyrus Leung
2025-07-27 19:42:40 -07:00
committed by GitHub
co-authored by Harry Mellor
parent 8f605ee309
commit 86ae693f20
94 changed files with 1117 additions and 1083 deletions
@@ -92,7 +92,7 @@ def _run_test(
# if we run HF first, the cuda initialization will be done and it
# will hurt multiprocessing backend with fork method (the default method).
with vllm_runner(model,
task="embed",
runner="pooling",
dtype=dtype,
enforce_eager=True,
max_model_len=8192) as vllm_model:
@@ -49,7 +49,7 @@ def vllm_reranker(
with vllm_runner(
model_name,
task="score",
runner="pooling",
dtype=dtype,
max_num_seqs=2,
max_model_len=2048,
@@ -64,7 +64,7 @@ def _run_test(
# if we run HF first, the cuda initialization will be done and it
# will hurt multiprocessing backend with fork method (the default method).
with vllm_runner(model,
task="embed",
runner="pooling",
dtype=dtype,
max_model_len=4096,
enforce_eager=True) as vllm_model:
@@ -44,7 +44,7 @@ def _run_test(
# vLLM needs a fresh new process without cuda initialization.
# if we run HF first, the cuda initialization will be done and it
# will hurt multiprocessing backend with fork method (the default method).
with vllm_runner(model, task="embed", dtype=dtype,
with vllm_runner(model, runner="pooling", dtype=dtype,
enforce_eager=True) as vllm_model:
vllm_outputs = vllm_model.embed(input_texts, images=input_images)
@@ -34,7 +34,7 @@ def _run_test(
set_default_torch_num_threads(1),
vllm_runner(
model,
task="embed",
runner="pooling",
dtype=torch.float16,
enforce_eager=True,
skip_tokenizer_init=True,