Skip to content

Add new deploy model cases 0426 #1392

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
kevalmorabia97 merged 6 commits into from
May 5, 2026
Merged

Add new deploy model cases 0426 #1392

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
8d2f99f
[Release-fix] Pin transformers<5.6 in release branch
kevalmorabia97 Apr 20, 2026
33d87c9
Add deploy cases for new NVIDIA models
Apr 26, 2026
c8cc706
Add deploy cases for new NVIDIA models
Apr 26, 2026
40e7045
Add deploy cases for Kimi Eagle3 and Qwen3-VL models
Apr 27, 2026
01baba7
Merge branch 'NVIDIA:main' into add-new-deploy-model-cases-0426
nvSiruiW May 5, 2026
0f45031
Apply suggestion from @kevalmorabia97
kevalmorabia97 May 5, 2026
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 pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,7 +82,7 @@ hf = [
"peft>=0.17.0",
"sentencepiece>=0.2.1", # Also implicitly used in test_unified_export_megatron, test_vllm_fakequant_megatron_export
"tiktoken",
"transformers>=4.56", # Should match modelopt/torch/__init__.py and noxfile.py
"transformers>=4.56,<5.6", # Should match modelopt/torch/__init__.py and noxfile.py
Comment thread
kevalmorabia97 marked this conversation as resolved.
Outdated
Comment thread
kevalmorabia97 marked this conversation as resolved.
Outdated
"wonderwords",
]

Expand Down
81 changes: 81 additions & 0 deletions tests/examples/llm_ptq/test_deploy.py
100644 → 100755
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -246,6 +246,12 @@ def test_llama(command):
tensor_parallel_size=4,
mini_sm=100,
),
*ModelDeployerList(
model_id="nvidia/Qwen3-VL-235B-A22B-Instruct-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
),
*ModelDeployerList(
model_id="nvidia/Qwen3-30B-A3B-NVFP4",
backend=("trtllm", "vllm", "sglang"),
Expand Down Expand Up @@ -346,6 +352,13 @@ def test_mixtral(command):
mini_sm=89,
attn_backend="FLASHINFER",
),
*ModelDeployerList(
model_id="nvidia/Gemma-4-31B-IT-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=1,
mini_sm=100,
attn_backend="FLASHINFER",
),
],
ids=idfn,
)
Expand Down Expand Up @@ -416,6 +429,44 @@ def test_kimi(command):
command.run()


@pytest.mark.parametrize(
"command",
[
*ModelDeployerList(
model_id="nvidia/GLM-4.7-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
),
*ModelDeployerList(
model_id="nvidia/GLM-5-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
),
],
ids=idfn,
)
def test_glm(command):
command.run()


@pytest.mark.parametrize(
"command",
[
*ModelDeployerList(
model_id="nvidia/MiniMax-M2.5-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
),
],
ids=idfn,
)
def test_minimax(command):
command.run()


@pytest.mark.parametrize(
"command",
[
Expand Down Expand Up @@ -451,6 +502,20 @@ def test_kimi(command):
mini_sm=89,
attn_backend="FLASHINFER",
),
*ModelDeployerList(
model_id="nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
attn_backend="FLASHINFER",
),
*ModelDeployerList(
model_id="nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8",
backend=("trtllm", "vllm", "sglang"),
tensor_parallel_size=8,
mini_sm=89,
attn_backend="FLASHINFER",
),
],
ids=idfn,
)
Expand Down Expand Up @@ -497,6 +562,22 @@ def test_medusa(command):
tensor_parallel_size=8,
mini_sm=89,
),
*ModelDeployerList(
base_model="nvidia/Kimi-K2-Thinking-NVFP4",
model_id="nvidia/Kimi-K2-Thinking-Eagle3",
backend=("trtllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
eagle3_one_model=False,
),
*ModelDeployerList(
base_model="nvidia/Kimi-K2.5-NVFP4",
model_id="nvidia/Kimi-K2.5-Thinking-Eagle3",
backend=("trtllm", "sglang"),
tensor_parallel_size=8,
mini_sm=100,
eagle3_one_model=False,
),
*ModelDeployerList(
base_model="Qwen/Qwen3-235B-A22B",
model_id="nvidia/Qwen3-235B-A22B-Eagle3",
Expand Down
Loading