Skip to content

Add GPT OSS tests - Vanilla GPT OSS, Speculative Decoding with EAGLE 3 and LoRA tests #2933

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

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Add GPT OSS tests - Vanilla GPT OSS, Speculative Decoding with EAGLE 3 and LoRA tests #2933

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
983a558
Add GPT OSS tests
Oct 31, 2025
bfe9569
Use gpt-oss from s3 bucket and GPT-OSS adapters and speculative_model
Nov 1, 2025
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
10 changes: 10 additions & 0 deletions tests/integration/llm/client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -338,6 +338,12 @@ def get_model_name():
"adapters": ["french", "spanish"],
"tokenizer": "unsloth/llama-3-8b-Instruct"
},
"gpt-oss-20b-lora": {
"batch_size": [4],
"seq_length": [16, 32],
"worker": 1,
"adapters": ["promptinj", "uncensored"],
},
"gemma-7b-unmerged-lora": {
"batch_size": [4],
"seq_length": [16, 32],
Expand Down Expand Up @@ -365,6 +371,10 @@ def get_model_name():
"batch_size": [1, 4],
"seq_length": [256],
},
"gpt-oss": {
"batch_size": [1, 4],
"seq_length": [256],
},
"tinyllama-input-len-exceeded": {
"batch_size": [1],
"seq_length": [25],
Expand Down
38 changes: 38 additions & 0 deletions tests/integration/llm/prepare.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -322,6 +322,11 @@
"option.task": "text-generation",
"option.tensor_parallel_degree": 4
},
"gpt-oss-20b": {
"option.model_id": "s3://djl-llm/gpt-oss-20b",
"option.task": "text-generation",
"option.tensor_parallel_degree": 4
},
"mistral-7b": {
"option.model_id": "s3://djl-llm/mistral-7b-instruct-v03",
"option.task": "text-generation",
Expand Down Expand Up @@ -372,6 +377,16 @@
"option.tensor_parallel_degree": 1,
"option.max_rolling_batch_size": 4,
},
"gpt-oss-20b-speculative-eagle3": {
"option.model_id": "s3://djl-llm/gpt-oss-20b",
"option.task": "text-generation",
"option.speculative_model": "zhuyksir/EAGLE3-gpt-oss-20b-bf16",
"option.speculative_method": "eagle3",
"option.num_speculative_tokens": 4,
"option.use_v2_block_manager": True,
"option.tensor_parallel_degree": 1,
"option.max_rolling_batch_size": 4,
},
"llama-68m-speculative-eagle": {
"option.model_id": "s3://djl-llm/llama-68m/",
"option.task": "text-generation",
Expand Down Expand Up @@ -554,6 +569,29 @@
"option.gpu_memory_utilization":
"0.8",
},
"gpt-oss-20b-unmerged-lora": {
"option.model_id":
"s3://djl-llm/gpt-oss-20b",
"option.tensor_parallel_degree":
"max",
"option.enable_lora":
"true",
"option.max_loras":
1,
"option.max_lora_rank":
128,
"option.long_lora_scaling_factors":
"4.0",
"option.adapters":
"adapters",
"adapter_ids": [
"waliboii/gpt-oss-20b-promptinj-lora",
"jworks/gpt-oss-20b-uncensored-lora",
],
"adapter_names": ["promptinj", "uncensored"],
"option.gpu_memory_utilization":
"0.8",
},
"phi2-unmerged-lora": {
"option.model_id":
"s3://djl-llm/phi-2/",
Expand Down
18 changes: 18 additions & 0 deletions tests/integration/tests.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -376,6 +376,12 @@ def test_llama2_7b_chat(self):
r.launch()
client.run("vllm_chat llama2-7b-chat".split())

def test_gpt_oss_20b(self):
with Runner('lmi', 'gpt-oss-20b') as r:
prepare.build_vllm_async_model("gpt-oss-20b")
r.launch()
client.run("vllm gpt-oss".split())

@pytest.mark.skipif(not is_applicable_cuda_capability(89),
reason="Unsupported CUDA capability")
def test_qwen2_7b_fp8(self):
Expand Down Expand Up @@ -409,6 +415,12 @@ def test_llama_68m_speculative_medusa(self):
@pytest.mark.gpu_4
class TestVllm2:

def test_gpt_oss_speculative_eagle3(self):
with Runner('lmi', 'gpt-oss-20b-speculative-eagle3') as r:
prepare.build_vllm_async_model("gpt-oss-20b-speculative-eagle3")
r.launch()
client.run("vllm gpt-oss".split())

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")
Expand Down Expand Up @@ -482,6 +494,12 @@ def test_lora_phi2(self):
@pytest.mark.gpu_4
class TestVllmAsyncLora:

def test_gpt_oss_20b_lora(self):
with Runner('lmi', 'gpt-oss-20b-unmerged-lora') as r:
prepare.build_vllm_async_model("gpt-oss-20b-unmerged-lora")
r.launch()
client.run("vllm_async_adapters gpt-oss-20b-lora".split())

def test_lora_llama3_8b_async(self):
with Runner('lmi', 'llama3-8b-unmerged-lora-async') as r:
prepare.build_vllm_async_model("llama3-8b-unmerged-lora")
Expand Down
Loading