mirror of
https://github.com/linkedin/Liger-Kernel.git
synced 2026-10-12 02:17:47 +00:00
Vaibhav JindalandCopilot
38c6a455ea
## Summary The three opt-in GPU DSL backends — **cutedsl**, **cutile**, and **cute** (fused-MoE) — had their tests inconsistently placed: a partial `test/cutedsl/` folder existed, but most DSL tests were scattered in `test/transformers/`. This PR gives each DSL its **own test folder**, mirroring the existing `test/cutedsl/` convention (drop the DSL prefix from filenames), and wires up matching `make` targets. One minor change in cron schedule for nvidia-ci. ## Changes ### cutedsl → `test/cutedsl/` - Moved `test_cutedsl_fused_linear_cross_entropy_sm90` → `test_fused_linear_cross_entropy_sm90` - Moved `test_cutedsl_rms_norm_fastpath` → `test_rms_norm_fastpath` - Moved `test_cutedsl_rope` → `test_rope` - Moved `test_swiglu_cutedsl` → `test_swiglu` - **Deleted** `test_cutedsl_rms_norm` — it was a duplicate *subset* of the canonical `test/cutedsl/test_rms_norm.py` (same two test-function names; the canonical file is a strict superset with the extra Blackwell fast-path tests and stricter mode-specific bf16 dW tolerances). Deleting avoids a filename + test-name collision; the canonical file is left untouched. - Fixed a stale docstring path in `test/cutedsl/test_backend.py`. ### cutile → new `test/cutile/` - Moved `test_cutile_backend` → `test_backend`, `test_cutile_fused_linear_cross_entropy_sm90` → `test_fused_linear_cross_entropy_sm90`, `test_cutile_rms_norm` → `test_rms_norm`, `test_cutile_rope` → `test_rope`, `test_cutile_swiglu` → `test_swiglu`. - **Split** the 7 env-gated `test_cutile_*` tests out of the shared `test/transformers/test_fused_linear_scaled_cross_entropy.py` into `test/cutile/test_fused_linear_scaled_cross_entropy.py`. The backend-agnostic frontend/dispatcher tests stay in `test/transformers/`. ### cute (fused-MoE) → new `test/cute/` - Moved `test_cute_moe_autograd` → `test_moe_autograd` and `test_moe` → `test_moe`. - `test_fused_moe.py` is the general Triton MoE test and **stays** in `test/transformers/`. ### Makefile - `test:` and `test-verbose:` now `--ignore=test/cutile` and `--ignore=test/cute` (and `test-verbose` also ignores `test/cutedsl` for consistency), so DSL tests run only via their dedicated targets. - New **`test-cutile`** (`LIGER_KERNEL_IMPL=cutile`, mirrors `test-cutedsl`). - New **`test-cute`** — intentionally does **not** set `LIGER_KERNEL_IMPL=cute`, because selecting it would force `import liger_kernel.ops` to load the separate `liger_cute_kernels` (lck) wheel and break the always-on registration/import-guard tests. ## Notes / decisions - All moves use `git mv` to preserve history. The `test.utils` import blocks in the moved rope/swiglu files were re-sorted by ruff because `test` is now first-party inside the packaged DSL folders. - **Coverage shift (intended):** a few always-on pure-Python tests (cutedsl rms_norm fastpath, cute registration/import-guard) previously ran under `make test`; they now run via the dedicated `test-*` targets, consistent with the one-folder-per-DSL model. - No new basename collisions: the DSL folders all have `__init__.py` (unique package-qualified module names) and are ignored by `make test`. ## Validation - `ruff check` and `ruff format --check` pass across the whole `test/` tree. - All affected files pass `py_compile`; `make -n` dry-runs verified for the new targets. - ⚠️ GPU/Modal test execution (`make test-cutedsl` / `test-cutile` / `test-cute`) was **not** run in this environment (no CUDA/torch available). These should be validated on a GPU host with the corresponding extras installed. --------- Co-authored-by: Copilot <[email protected]>
107 lines
3.5 KiB
Makefile
107 lines
3.5 KiB
Makefile
.PHONY: test test-verbose checkstyle test-convergence test-cutedsl test-cutile test-cute all serve build clean
|
|
|
|
|
|
all: checkstyle test test-convergence
|
|
|
|
# Command to run pytest for correctness tests
|
|
test:
|
|
python -m pytest --disable-warnings \
|
|
--ignore=test/convergence \
|
|
--ignore=test/cutedsl \
|
|
--ignore=test/cutile \
|
|
--ignore=test/cute \
|
|
test/
|
|
|
|
# Command to run pytest for CuTe DSL kernel tests
|
|
# Requires nvidia-cutlass-dsl: pip install -e '.[cutedsl]'
|
|
test-cutedsl:
|
|
LIGER_KERNEL_IMPL=cutedsl python -m pytest --disable-warnings test/cutedsl/
|
|
|
|
# Command to run pytest for cuTile kernel tests
|
|
# Requires cuda-tile: pip install -e '.[cutile]' (or '.[cutile-tileiras]')
|
|
test-cutile:
|
|
LIGER_KERNEL_IMPL=cutile python -m pytest --disable-warnings test/cutile/
|
|
|
|
# Command to run pytest for the cute fused-MoE tests
|
|
# The pure-Python registration/import-guard tests always run; the functional
|
|
# tests need the separate liger_cute_kernels (lck) wheel and >=2 CUDA devices.
|
|
# LIGER_KERNEL_IMPL is intentionally NOT set to 'cute': selecting it would force
|
|
# `import liger_kernel.ops` to load the lck wheel and break the always-on tests.
|
|
test-cute:
|
|
python -m pytest --disable-warnings test/cute/
|
|
|
|
# Command to run pytest for correctness tests with coverage reporting and full test-duration breakdown
|
|
test-verbose:
|
|
python -m pytest --disable-warnings \
|
|
--cov=src/liger_kernel \
|
|
--cov-report=term-missing \
|
|
--cov-report=html \
|
|
--cov-config=pyproject.toml \
|
|
--durations=0 \
|
|
--ignore=test/convergence \
|
|
--ignore=test/cutedsl \
|
|
--ignore=test/cutile \
|
|
--ignore=test/cute \
|
|
test/
|
|
|
|
# Command to run coverage report
|
|
coverage:
|
|
coverage report -m
|
|
|
|
# Command to run ruff for linting and formatting code
|
|
checkstyle:
|
|
ruff check --output-format=concise .; ruff_check_status=$$?; \
|
|
ruff format --check --diff .; ruff_format_status=$$?; \
|
|
ruff check . --fix; \
|
|
ruff format .; \
|
|
if [ $$ruff_check_status -ne 0 ] || [ $$ruff_format_status -ne 0 ]; then \
|
|
exit 1; \
|
|
fi
|
|
|
|
# Command to run pytest for convergence tests
|
|
# We have to explicitly set HF_DATASETS_OFFLINE=1, or dataset will silently try to send metrics and timeout (80s) https://github.com/huggingface/datasets/blob/37a603679f451826cfafd8aae00738b01dcb9d58/src/datasets/load.py#L286
|
|
test-convergence:
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings \
|
|
test/convergence/fp32/test_mini_models.py \
|
|
test/convergence/fp32/test_mini_models_multimodal.py \
|
|
test/convergence/fp32/test_mini_models_with_logits.py \
|
|
test/convergence/bf16/test_mini_models.py \
|
|
test/convergence/bf16/test_mini_models_multimodal.py \
|
|
test/convergence/bf16/test_mini_models_with_logits.py
|
|
|
|
# Command to run all benchmark scripts and update benchmarking data file
|
|
# By default this doesn't overwrite existing data for the same benchmark experiment
|
|
# run with `make run-benchmarks OVERWRITE=1` to overwrite existing benchmark data
|
|
BENCHMARK_DIR = benchmark/scripts
|
|
BENCHMARK_SCRIPTS = $(wildcard $(BENCHMARK_DIR)/benchmark_*.py)
|
|
OVERWRITE ?= 0
|
|
|
|
run-benchmarks:
|
|
@for script in $(BENCHMARK_SCRIPTS); do \
|
|
echo "Running benchmark: $$script"; \
|
|
if [ $(OVERWRITE) -eq 1 ]; then \
|
|
python $$script --overwrite; \
|
|
else \
|
|
python $$script; \
|
|
fi; \
|
|
done
|
|
|
|
# MkDocs Configuration
|
|
MKDOCS = mkdocs
|
|
CONFIG_FILE = mkdocs.yml
|
|
SITE_DIR = site
|
|
|
|
# MkDocs targets
|
|
|
|
# Serve the documentation
|
|
serve:
|
|
$(MKDOCS) serve -f $(CONFIG_FILE)
|
|
|
|
# Build the documentation into the specified site directory
|
|
build:
|
|
$(MKDOCS) build -f $(CONFIG_FILE) --site-dir $(SITE_DIR)
|
|
|
|
# Clean the output directory
|
|
clean:
|
|
rm -rf $(SITE_DIR)/
|