mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
chore : split GptExecutor tests out of gpt tests to reduce single test time (#3412)
Signed-off-by: peaceh <103117813+peaceh-nv@users.noreply.github.com> Co-authored-by: QI JUN <22017000+QiJune@users.noreply.github.com>
This commit is contained in:
parent
8d164f40d7
commit
215fb20567
@ -20,6 +20,7 @@ default_test_timeout = 3600
|
||||
|
||||
include_test_map = {
|
||||
"gpt": ("Gpt[^j]", ),
|
||||
"gpt_executor": ("GptExecutor", ),
|
||||
"gptj": ("Gptj", ),
|
||||
"llama": ("Llama", ),
|
||||
"chatglm": ("ChatGlm", ),
|
||||
@ -40,6 +41,7 @@ include_test_map = {
|
||||
|
||||
def generate_excluded_model_tests() -> Generator[str, None, None]:
|
||||
yield "Gpt[^j]"
|
||||
yield "GptExecutor"
|
||||
yield "Gptj"
|
||||
yield "Llama"
|
||||
yield "ChatGlm"
|
||||
@ -619,6 +621,10 @@ def prepare_model_tests(model_name: str,
|
||||
beams_arg = ['--beams', '1,2']
|
||||
model_name = 'enc_dec'
|
||||
|
||||
# share the same script for gpt and gpt_executor
|
||||
if model_name == 'gpt_executor':
|
||||
model_name = 'gpt'
|
||||
|
||||
build_engines = [
|
||||
python_exe,
|
||||
str(scripts_dir / f"build_{model_name}_engines.py")
|
||||
@ -710,6 +716,9 @@ def run_single_gpu_tests(build_dir: _pl.Path,
|
||||
|
||||
excluded_tests = ["FP8"] if not run_fp8 else []
|
||||
|
||||
if "gpt" in test_list and "gpt_executor" not in test_list:
|
||||
excluded_tests.append("GptExecutor")
|
||||
|
||||
ctest = ["ctest", "--output-on-failure", "--output-junit", resultFileName]
|
||||
|
||||
if included_tests:
|
||||
|
||||
@ -335,7 +335,8 @@ def test_unit_tests(build_google_tests, build_dir, lora_setup):
|
||||
indirect=True)
|
||||
@pytest.mark.parametrize("model", [
|
||||
"bart", "chatglm", "eagle", "encoder", "enc_dec_language_adapter", "gpt",
|
||||
"llama", "mamba", "medusa", "recurrentgemma", "redrafter", "t5"
|
||||
"gpt_executor", "llama", "mamba", "medusa", "recurrentgemma", "redrafter",
|
||||
"t5"
|
||||
])
|
||||
@pytest.mark.parametrize("run_fp8", [False, True], ids=["", "fp8"])
|
||||
def test_model(build_google_tests, model, prepare_model, run_model_tests,
|
||||
|
||||
@ -40,6 +40,7 @@ l0_a30:
|
||||
# ------------- CPP tests ---------------
|
||||
- test_cpp.py::test_unit_tests[80]
|
||||
- test_cpp.py::test_model[gpt-80]
|
||||
- test_cpp.py::test_model[gpt_executor-80]
|
||||
- test_cpp.py::test_benchmarks[gpt-80]
|
||||
- condition:
|
||||
ranges:
|
||||
|
||||
Loading…
Reference in New Issue
Block a user