Skip to content

feat: Add Phi-4-Mini-Instruct in Pytorch backend for LLM API accuracy tests #6303

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
schetlur-nv merged 2 commits into from
Jul 28, 2025
Merged

feat: Add Phi-4-Mini-Instruct in Pytorch backend for LLM API accuracy tests #6303

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
8728706
Update the test
moraxu Jul 23, 2025
0f931a2
Update the QA test lists
moraxu Jul 23, 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
2 changes: 2 additions & 0 deletions tests/integration/defs/accuracy/references/cnn_dailymail.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,8 @@ microsoft/Phi-3-small-128k-instruct:
- accuracy: 27.208
microsoft/Phi-3.5-mini-instruct:
- accuracy: 31.354
microsoft/Phi-4-mini-instruct:
- accuracy: 32.921
state-spaces/mamba-130m-hf:
- accuracy: 19.470
lmsys/vicuna-7b-v1.3:
Expand Down
2 changes: 2 additions & 0 deletions tests/integration/defs/accuracy/references/gsm8k.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,3 +124,5 @@ mistralai/Ministral-8B-Instruct-2410:
accuracy: 78.35
microsoft/Phi-4-multimodal-instruct:
- accuracy: 81.19
microsoft/Phi-4-mini-instruct:
- accuracy: 82.30
7 changes: 0 additions & 7 deletions tests/integration/defs/accuracy/test_llm_api_pytorch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1872,10 +1872,6 @@ class TestPhi4MiniInstruct(LlmapiAccuracyTestHarness):
MODEL_NAME = "microsoft/Phi-4-mini-instruct"
MODEL_PATH = f"{llm_models_root()}/Phi-4-mini-instruct"

@pytest.mark.skip(
reason=
"Temporarily skipping test_auto_dtype while resolving Phi-4's architecture issue."
)
def test_auto_dtype(self):
with LLM(self.MODEL_PATH) as llm:
task = CnnDailymail(self.MODEL_NAME)
Expand All @@ -1884,9 +1880,6 @@ def test_auto_dtype(self):
task.evaluate(llm)
task = GSM8K(self.MODEL_NAME)
task.evaluate(llm)
task = GPQADiamond(self.MODEL_NAME)
task.evaluate(llm,
extra_evaluator_kwargs=dict(apply_chat_template=True))


class TestKanana_Instruct(LlmapiAccuracyTestHarness):
Expand Down
1 change: 1 addition & 0 deletions tests/integration/test_lists/qa/examples_test_list.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -491,6 +491,7 @@ accuracy/test_llm_api_pytorch.py::TestBielik11BInstruct::test_fp8
accuracy/test_llm_api_pytorch.py::TestMinistral8BInstruct::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestMinistral8BInstruct::test_fp8
accuracy/test_llm_api_pytorch.py::TestPhi4MM::test_auto_dtype
accuracy/test_llm_api_pytorch.py::TestPhi4MiniInstruct::test_auto_dtype

test_e2e.py::test_llama_e2e[use_cpp_session-remove_input_padding-]
test_e2e.py::test_llama_e2e[use_py_session-remove_input_padding-]
Expand Down
1 change: 1 addition & 0 deletions tests/integration/test_lists/qa/llm_sanity_test.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,6 +54,7 @@ accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_fp8_block_scales[laten
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[latency_moe_cutlass]
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_nvfp4[latency_moe_trtllm]
accuracy/test_llm_api_pytorch.py::TestQwen3_8B::test_fp8_block_scales[latency]
accuracy/test_llm_api_pytorch.py::TestPhi4MiniInstruct::test_auto_dtype
disaggregated/test_disaggregated.py::test_disaggregated_cache_aware_balance[TinyLlama-1.1B-Chat-v1.0]
disaggregated/test_disaggregated.py::test_disaggregated_cuda_graph[TinyLlama-1.1B-Chat-v1.0]
disaggregated/test_disaggregated.py::test_disaggregated_deepseek_v3_lite_fp8_attention_dp_one_mtp[DeepSeek-V3-Lite-fp8]
Expand Down