Skip to content

[ci] Fix vllm spec decoding config integration test #2938

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
xyang16 merged 1 commit into from
Nov 3, 2025
Merged

[ci] Fix vllm spec decoding config integration test #2938

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion serving/docs/lmi/user_guides/vllm_user_guide.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,7 +182,7 @@ These are supported in LMI.
For example, if you want to enable the `speculative_config`, you can do:

* `option.speculative_config={"model": "meta-llama/Llama3.2-1B-Instruct", "num_speculative_tokens": 5}`
* `OPTION_SPECULATIVE_CONFIG={"model": "meta-llama/Llama3.2-1B-Instruct", "num_speculative_tokens": 5}`
* `OPTION_SPECULATIVE_CONFIG='{"model": "meta-llama/Llama3.2-1B-Instruct", "num_speculative_tokens": 5}'`


## Custom Handlers
Expand Down
4 changes: 2 additions & 2 deletions tests/integration/llm/client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -284,10 +284,10 @@ def get_model_name():
"seq_length": [256],
"tokenizer": "JackFram/llama-68m"
},
"llama-68m-speculative-eagle": {
"llama3-1-8b-speculative-eagle": {
"batch_size": [1, 4],
"seq_length": [256],
"tokenizer": "JackFram/llama-68m"
"tokenizer": "unsloth/Meta-Llama-3.1-8B"
},
"llama-7b-unmerged-lora": {
"batch_size": [3],
Expand Down
16 changes: 7 additions & 9 deletions tests/integration/llm/prepare.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -366,19 +366,17 @@
"llama-68m-speculative-medusa": {
"option.model_id": "s3://djl-llm/llama-68m/",
"option.task": "text-generation",
"option.speculative_model": "abhigoyal/vllm-medusa-llama-68m-random",
"option.num_speculative_tokens": 4,
"option.use_v2_block_manager": True,
"option.speculative_config":
'{"method":"medusa","model":"abhigoyal/vllm-medusa-llama-68m-random","num_speculative_tokens":4}',
"option.tensor_parallel_degree": 1,
"option.max_rolling_batch_size": 4,
},
"llama-68m-speculative-eagle": {
"option.model_id": "s3://djl-llm/llama-68m/",
"llama3-1-8b-speculative-eagle": {
"option.model_id": "s3://djl-llm/llama-3.1-8b-hf/",
"option.task": "text-generation",
"option.speculative_model": "abhigoyal/vllm-eagle-llama-68m-random",
"option.num_speculative_tokens": 4,
"option.use_v2_block_manager": True,
"option.tensor_parallel_degree": 1,
"option.speculative_config":
'{"method":"eagle","model":"yuhuili/EAGLE-LLaMA3.1-Instruct-8B","num_speculative_tokens":4}',
"option.tensor_parallel_degree": 4,
"option.max_rolling_batch_size": 4,
},
"llama-7b-unmerged-lora": {
Expand Down
8 changes: 4 additions & 4 deletions tests/integration/tests.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -409,11 +409,11 @@ def test_llama_68m_speculative_medusa(self):
@pytest.mark.gpu_4
class TestVllm2:

def test_llama_68m_speculative_eagle(self):
with Runner('lmi', 'llama-68m-speculative-eagle') as r:
prepare.build_vllm_async_model("llama-68m-speculative-eagle")
def test_llama3_1_8b_speculative_eagle(self):
with Runner('lmi', 'llama3-1-8b-speculative-eagle') as r:
prepare.build_vllm_async_model("llama3-1-8b-speculative-eagle")
r.launch()
client.run("vllm llama-68m-speculative-eagle".split())
client.run("vllm llama3-1-8b-speculative-eagle".split())

def test_llama3_1_8b_instruct_tool(self):
with Runner('lmi', 'llama3-1-8b-instruct-tool') as r:
Expand Down
Loading