Add pydocstyle to the ruff rules (#52136)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Signed-off-by: Harry Mellor <hej.mellor@gmail.com> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5
parent
804e5377aa
commit
8c1557a79c
@@ -45,7 +45,8 @@ def _bgmv_shrink(
|
||||
inputs, lora_weight, output, seq_len_tensor, lora_indices, scaling=1.0
|
||||
):
|
||||
"""Memory-efficient shrink reference: per-LoRA matmul loop.
|
||||
output[mask] = scaling * inputs[mask] @ weight.T"""
|
||||
output[mask] = scaling * inputs[mask] @ weight.T
|
||||
"""
|
||||
exploded = torch.repeat_interleave(lora_indices, seq_len_tensor)
|
||||
for lid in exploded.unique():
|
||||
if lid < 0:
|
||||
@@ -66,7 +67,8 @@ def _bgmv_expand(
|
||||
add_inputs=False,
|
||||
):
|
||||
"""Memory-efficient expand reference: per-LoRA matmul loop.
|
||||
output[mask, offset:offset+n] (+)= inputs[mask] @ weight.T"""
|
||||
output[mask, offset:offset+n] (+)= inputs[mask] @ weight.T
|
||||
"""
|
||||
exploded = torch.repeat_interleave(lora_indices, seq_len_tensor)
|
||||
for lid in exploded.unique():
|
||||
if lid < 0:
|
||||
@@ -160,8 +162,7 @@ def check_lora_shrink_kernel(
|
||||
seq_length: int,
|
||||
scaling: float,
|
||||
):
|
||||
"""
|
||||
Compare outputs of torch_ops.sgmv_shrink and triton_ops.lora_shrink
|
||||
"""Compare outputs of torch_ops.sgmv_shrink and triton_ops.lora_shrink
|
||||
kernels.
|
||||
"""
|
||||
data: PunicaTensors = generate_data_for_nslices(
|
||||
@@ -234,8 +235,7 @@ def check_lora_expand_kernel(
|
||||
seq_length: int,
|
||||
add_inputs: bool,
|
||||
):
|
||||
"""
|
||||
Compare outputs of torch_ops.sgmv_expand and triton_ops.lora_expand
|
||||
"""Compare outputs of torch_ops.sgmv_expand and triton_ops.lora_expand
|
||||
kernels.
|
||||
"""
|
||||
data: PunicaTensors = generate_data_for_nslices(
|
||||
@@ -438,9 +438,7 @@ def test_kernels(
|
||||
seed: int,
|
||||
op_type: str,
|
||||
):
|
||||
"""
|
||||
Tests LoRA kernels.
|
||||
"""
|
||||
"""Tests LoRA kernels."""
|
||||
torch.set_default_device(device)
|
||||
torch.accelerator.set_device_index(device)
|
||||
set_random_seed(seed)
|
||||
@@ -491,9 +489,7 @@ def test_kernels_hidden_size(
|
||||
seed: int,
|
||||
op_type: str,
|
||||
):
|
||||
"""
|
||||
Tests SGMV and LoRA kernels.
|
||||
"""
|
||||
"""Tests SGMV and LoRA kernels."""
|
||||
torch.set_default_device(device)
|
||||
torch.accelerator.set_device_index(device)
|
||||
set_random_seed(seed)
|
||||
@@ -526,8 +522,7 @@ def test_kernels_hidden_size(
|
||||
|
||||
@pytest.mark.parametrize("device", DEVICES)
|
||||
def test_add_lora_fused_moe_early_exit(device):
|
||||
"""
|
||||
Ensures add_lora_fused_moe does not invoke the LoRA kernel or
|
||||
"""Ensures add_lora_fused_moe does not invoke the LoRA kernel or
|
||||
modify the output tensor when no_lora_flag_cpu is True
|
||||
"""
|
||||
from types import SimpleNamespace
|
||||
|
||||
Reference in New Issue
Block a user