[Refactor] Pass Renderer to Input Processor (#34329)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
@@ -507,7 +507,8 @@ def test_apc_single_prompt_block_align_alignment(
|
||||
vllm_runner_kwargs["enable_prefix_caching"] = True
|
||||
with vllm_runner(**vllm_runner_kwargs) as vllm_model:
|
||||
# Retrieve the default mamba state block size
|
||||
mamba_block_size = vllm_model.llm.llm_engine.cache_config.mamba_block_size
|
||||
vllm_config = vllm_model.llm.llm_engine.vllm_config
|
||||
mamba_block_size = vllm_config.cache_config.mamba_block_size
|
||||
|
||||
# In case the hybrid model does not have the
|
||||
# "mamba_block_size" assume a fixed constant
|
||||
@@ -660,7 +661,8 @@ def test_apc_multiple_prompts_block_align_alignment(
|
||||
vllm_runner_kwargs["enable_prefix_caching"] = True
|
||||
with vllm_runner(**vllm_runner_kwargs) as vllm_model:
|
||||
# Retrieve the default mamba state block size
|
||||
mamba_block_size = vllm_model.llm.llm_engine.cache_config.mamba_block_size
|
||||
vllm_config = vllm_model.llm.llm_engine.vllm_config
|
||||
mamba_block_size = vllm_config.cache_config.mamba_block_size
|
||||
|
||||
# In case the hybrid model does not have the
|
||||
# "mamba_block_size" assume a fixed constant
|
||||
|
||||
@@ -25,7 +25,8 @@ def test_classify_models(
|
||||
with vllm_runner(
|
||||
model, max_model_len=512, dtype=dtype, enable_prefix_caching=True
|
||||
) as vllm_model:
|
||||
cache_config = vllm_model.llm.llm_engine.cache_config
|
||||
vllm_config = vllm_model.llm.llm_engine.vllm_config
|
||||
cache_config = vllm_config.cache_config
|
||||
assert cache_config.enable_prefix_caching
|
||||
|
||||
# First Run
|
||||
@@ -74,7 +75,8 @@ def test_embed_models(
|
||||
max_model_len=None,
|
||||
enable_prefix_caching=True,
|
||||
) as vllm_model:
|
||||
cache_config = vllm_model.llm.llm_engine.cache_config
|
||||
vllm_config = vllm_model.llm.llm_engine.vllm_config
|
||||
cache_config = vllm_config.cache_config
|
||||
assert cache_config.enable_prefix_caching
|
||||
|
||||
# First Run
|
||||
@@ -106,5 +108,6 @@ def test_non_causal_models(
|
||||
hf_runner, vllm_runner, example_prompts, model: str, dtype: str
|
||||
) -> None:
|
||||
with vllm_runner(model, max_model_len=512, dtype=dtype) as vllm_model:
|
||||
cache_config = vllm_model.llm.llm_engine.cache_config
|
||||
vllm_config = vllm_model.llm.llm_engine.vllm_config
|
||||
cache_config = vllm_config.cache_config
|
||||
assert not cache_config.enable_prefix_caching
|
||||
|
||||
Reference in New Issue
Block a user