mirror of
https://github.com/linkedin/Liger-Kernel.git
synced 2026-10-12 02:17:47 +00:00
jiahao su
1fdb668d35
Summary
----------
1. I found that the benchmark_fused_add_rms_norm.py format issue led to
the collective failure of CI. fix it first.
2. The checkstyle script has been updated to enable it to display the
modified logs.
Test
```
>> make 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
All checks passed!
--- benchmark/scripts/benchmark_fused_add_rms_norm.py
+++ benchmark/scripts/benchmark_fused_add_rms_norm.py
@@ -103,7 +103,7 @@
elif mode == "backward":
y, s = y_fwd()
ms_50, ms_20, ms_80 = triton.testing.do_bench(
- lambda: (torch.autograd.backward((y, s), (dy, ds), retain_graph=True)),
+ lambda: torch.autograd.backward((y, s), (dy, ds), retain_graph=True),
grad_to_none=[x, r],
rep=500,
quantiles=QUANTILES,
1 file would be reformatted, 226 files already formatted
All checks passed!
1 file reformatted, 226 files left unchanged
make: *** [checkstyle] Error 1
```
74 lines
2.5 KiB
Makefile
74 lines
2.5 KiB
Makefile
.PHONY: test checkstyle test-convergence all serve build clean
|
|
|
|
|
|
all: checkstyle test test-convergence
|
|
|
|
# Command to run pytest for correctness tests
|
|
test:
|
|
python -m pytest --disable-warnings \
|
|
--cov=src/liger_kernel \
|
|
--cov-report=term-missing \
|
|
--ignore=test/convergence \
|
|
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
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings test/convergence/fp32/test_mini_models_multimodal.py
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings test/convergence/fp32/test_mini_models_with_logits.py
|
|
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings test/convergence/bf16/test_mini_models.py
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings test/convergence/bf16/test_mini_models_multimodal.py
|
|
HF_DATASETS_OFFLINE=1 python -m pytest --disable-warnings 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)/
|