Skip to content

GitLab

Unverified Commit 1575c170 authored by Jee Jee Li's avatar Jee Jee Li Committed by GitHub
Browse files

[CI/Build] Fix LoRA OOM (#16624) 


Signed-off-by: default avatarJee Jee Li <pandaleefree@gmail.com>
parent 6ae996a8
Hide whitespace changes
Inline Side-by-side
Showing with 13 additions and 4 deletions
tests/lora/test_minicpmv_tp.py
View file @ 1575c170
...@@ -66,8 +66,12 @@ def test_minicpmv_lora(minicpmv_lora_files): ...@@ -66,8 +66,12 @@ def test_minicpmv_lora(minicpmv_lora_files):
max_loras=2, max_loras=2,
max_lora_rank=8, max_lora_rank=8,
enforce_eager=True, enforce_eager=True,
max_model_len=2048,
limit_mm_per_prompt={
"image": 2,
"video": 0
},
trust_remote_code=True, trust_remote_code=True,
enable_chunked_prefill=True,
) )
output1 = do_sample(llm, minicpmv_lora_files, lora_id=1) output1 = do_sample(llm, minicpmv_lora_files, lora_id=1)
for i in range(len(EXPECTED_OUTPUT)): for i in range(len(EXPECTED_OUTPUT)):
...@@ -91,9 +95,11 @@ def test_minicpmv_tp4_wo_fully_sharded_loras(minicpmv_lora_files): ...@@ -91,9 +95,11 @@ def test_minicpmv_tp4_wo_fully_sharded_loras(minicpmv_lora_files):
max_loras=4, max_loras=4,
max_lora_rank=64, max_lora_rank=64,
tensor_parallel_size=4, tensor_parallel_size=4,
limit_mm_per_prompt={
"image": 2,
"video": 0
},
trust_remote_code=True, trust_remote_code=True,
enforce_eager=True,
enable_chunked_prefill=True,
) )
output_tp = do_sample(llm, minicpmv_lora_files, lora_id=1) output_tp = do_sample(llm, minicpmv_lora_files, lora_id=1)
for i in range(len(EXPECTED_OUTPUT)): for i in range(len(EXPECTED_OUTPUT)):
...@@ -115,8 +121,11 @@ def test_minicpmv_tp4_fully_sharded_loras(minicpmv_lora_files): ...@@ -115,8 +121,11 @@ def test_minicpmv_tp4_fully_sharded_loras(minicpmv_lora_files):
max_lora_rank=8, max_lora_rank=8,
tensor_parallel_size=4, tensor_parallel_size=4,
trust_remote_code=True, trust_remote_code=True,
limit_mm_per_prompt={
"image": 1,
"video": 0
},
fully_sharded_loras=True, fully_sharded_loras=True,
enable_chunked_prefill=True,
) )
output_tp = do_sample(llm, minicpmv_lora_files, lora_id=1) output_tp = do_sample(llm, minicpmv_lora_files, lora_id=1)
for i in range(len(EXPECTED_OUTPUT)): for i in range(len(EXPECTED_OUTPUT)):
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment