instruction stringclasses 100
values | code stringlengths 78 193k | response stringlengths 259 170k | file stringlengths 59 203 |
|---|---|---|---|
Generate docstrings for this script | import glob
import os
from typing import Dict, List, Union
import torch
from diffusers.utils import is_safetensors_available
from huggingface_hub.constants import HF_HUB_CACHE
if is_safetensors_available():
import safetensors.torch
from huggingface_hub import snapshot_download
from diffusers import DiffusionPi... | --- +++ @@ -22,6 +22,25 @@
class CheckpointMergerPipeline(DiffusionPipeline):
+ """
+ A class that supports merging diffusion models based on the discussion here:
+ https://github.com/huggingface/diffusers/issues/877
+
+ Example usage:-
+
+ pipe = DiffusionPipeline.from_pretrained("CompVis/stable-dif... | https://raw.githubusercontent.com/Sanster/IOPaint/HEAD/scripts/tool.py |
Create documentation for each function signature | # Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
import warnings
import numpy as np
import torch
from PIL import Image
def get_sdpa_settings():
if torch.cuda.is_ava... | --- +++ @@ -42,6 +42,15 @@
def mask_to_box(masks: torch.Tensor):
+ """
+ compute bounding box given an input mask
+
+ Inputs:
+ - masks: [B, 1, H, W] boxes, dtype=torch.Tensor
+
+ Returns:
+ - box_coords: [B, 1, 4], contains (x, y) coordinates of top left and bottom right box corners, dtype=torch.... | https://raw.githubusercontent.com/Sanster/IOPaint/HEAD/iopaint/plugins/segment_anything2/utils/misc.py |
Generate docstrings with parameter types | # Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
import torch
import torch.nn as nn
from torchvision.transforms import Normalize, Resize, ToTensor
class SAM2Transforms(n... | --- +++ @@ -13,6 +13,9 @@ def __init__(
self, resolution, mask_threshold, max_hole_area=0.0, max_sprinkle_area=0.0
):
+ """
+ Transforms for SAM2.
+ """
super().__init__()
self.resolution = resolution
self.mask_threshold = mask_threshold
@@ -40,6 +43,13 ... | https://raw.githubusercontent.com/Sanster/IOPaint/HEAD/iopaint/plugins/segment_anything2/utils/transforms.py |
Add minimal docstrings for each function |
import os
import sys
import re
import ctypes
import subprocess
from pathlib import Path
import cutlass
CUTE_DSL_PTXAS_PATH = os.environ.get("CUTE_DSL_PTXAS_PATH", None)
VERBOSE = os.environ.get("CUTE_DSL_PTXAS_VERBOSE", "0") == "1"
_original_load_cuda_library = None
_user_wanted_ptx = False # True if user origina... | --- +++ @@ -1,3 +1,9 @@+"""
+System ptxas replacement for CUTLASS DSL.
+Environment variables:
+ CUTE_DSL_PTXAS_PATH - Path to ptxas (e.g., /usr/local/cuda/bin/ptxas)
+ CUTE_DSL_PTXAS_VERBOSE - Set to 1 for verbose output
+"""
import os
import sys
@@ -22,6 +28,7 @@
def _get_ptx(compiled_func) -> tuple[... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/cute_dsl_ptxas.py |
Create docstrings for each class method | #!/usr/bin/env python
import argparse
import time
import torch
import torch.nn.functional as F
from flash_attn.cute.interface import _flash_attn_fwd, _flash_attn_bwd
# ── Helpers ────────────────────────────────────────────────────────────────
def parse_int_k(s):
s = s.strip().lower()
if s.endswith("k"):
... | --- +++ @@ -1,4 +1,37 @@ #!/usr/bin/env python
+"""Unified SM90 benchmark for forward and backward passes.
+
+Usage:
+ # Default: bench fwd+bwd for hdim 64,96,128 at seqlen 8192
+ python benchmarks/bench_sm90.py
+
+ # Forward only, specific hdims
+ python benchmarks/bench_sm90.py --direction fwd --hdim 64,9... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/benchmarks/bench_sm90.py |
Auto-generate documentation strings for this file | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# A reimplementation of https://github.com/Dao-AILab/flash-attention/blob/main/hopper/flash_bwd_postprocess_kernel.h
# from Cutlass C++ to Cute-DSL.
import math
from typing import Callable, Optional, Type
import cuda.... | --- +++ @@ -44,6 +44,12 @@ use_2cta_instrs: bool = False,
cluster_size: int = 1, # for varlen offsets
):
+ """
+ :param head_dim: head dimension
+ :type head_dim: int
+ :param tile_m: m block size
+ :type tile_m: int
+ """
self.dtype = dtype
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_bwd_postprocess.py |
Write reusable docstrings | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
from typing import Tuple, Optional
from dataclasses import dataclass
import cutlass
import cutlass.cute as cute
from cutlass import Int32, const_expr
from flash_attn.cute.seqlen_info import SeqlenInfoQK, SeqlenInfoQK... | --- +++ @@ -78,6 +78,11 @@ split_idx: Int32 = 0,
num_splits: Int32 = 1,
) -> Tuple[Int32, Int32]:
+ """Get the block range for new K tokens (append KV).
+
+ First computes the full n_block range via get_n_block_min_max, then maps
+ those blocks into the new-K index space by su... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/block_info.py |
Add clean documentation to messy code |
import math
import torch
try:
import cudnn
except ImportError:
cudnn = None
# ── FLOPS calculation ────────────────────────────────────────────────────────
def flops(
batch, nheads, seqlen_q, seqlen_k, headdim, headdim_v, causal=False, window_size=(None, None)
):
if causal:
avg_seqlen = (m... | --- +++ @@ -1,3 +1,4 @@+"""Shared benchmark utilities: attention_ref, cuDNN helpers, flops calculation."""
import math
import torch
@@ -43,6 +44,12 @@
def attention_ref(q, k, v, causal=False):
+ """Standard attention reference implementation.
+
+ Args:
+ q, k, v: (batch, seqlen, nheads, headdim) te... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/bench_utils.py |
Expand my code with proper documentation strings | # Copyright (c) 2025, Tri Dao.
import logging
import os
import sys
import cutlass.cute as cute
from cutlass import const_expr
_LOG_LEVEL_NAMES = {"off": 0, "host": 1, "kernel": 2, "max": 3}
def _parse_log_level(raw: str) -> int:
if raw in _LOG_LEVEL_NAMES:
return _LOG_LEVEL_NAMES[raw]
try:
... | --- +++ @@ -1,5 +1,29 @@ # Copyright (c) 2025, Tri Dao.
+"""Unified FlashAttention logging controlled by a single ``FA_LOG_LEVEL`` env var.
+
+Host-side messages go through Python ``logging`` (logger name ``flash_attn``).
+A default ``StreamHandler`` is attached automatically when ``FA_LOG_LEVEL >= 1``
+so that stand... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/fa_logging.py |
Annotate my code with docstrings |
from typing import Callable, NamedTuple, Tuple
import cutlass.cute as cute
import torch
from flash_attn.cute.cute_dsl_utils import get_broadcast_dims, to_cute_tensor
def ceildiv(a: int, b: int) -> int:
return (a + b - 1) // b
class BlockSparseTensors(NamedTuple):
mask_block_cnt: cute.Tensor
mask_bloc... | --- +++ @@ -1,3 +1,6 @@+"""
+Block-sparsity utilities for FlexAttention
+"""
from typing import Callable, NamedTuple, Tuple
@@ -38,6 +41,7 @@ context: str | None,
hint: str | Callable[[], str] | None,
) -> torch.Tensor:
+ """Check if we need to expand the tensor to expected shape, and do so if possibl... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/block_sparsity.py |
Write docstrings describing functionality | # Copyright (c) 2025, Tri Dao.
import os
import pathlib
from typing import Tuple
from functools import partial, lru_cache
from dataclasses import dataclass, fields
import torch
try:
from triton.tools.disasm import extract
except ImportError:
extract = None
import cutlass
import cutlass.cute as cute
from cut... | --- +++ @@ -85,6 +85,7 @@
def cute_compile_patched(*args, **kwargs):
+ """A patched version of cute.compile that dump the SASS to a file if CUTE_CUBIN_PATH is set."""
cubin_path = os.getenv("CUTE_CUBIN_PATH", None)
if cubin_path is not None:
cutlass.base_dsl.runtime.cuda.load_cubin_module_data... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/cute_dsl_utils.py |
Write docstrings for utility functions |
from typing import Callable, Optional
from functools import partial
import math
import cutlass
import cutlass.cute as cute
from cutlass import Float32, Int32, const_expr
from quack import copy_utils
# Import data structures from block_sparsity
from flash_attn.cute.block_sparsity import BlockSparseTensors
from flash_... | --- +++ @@ -1,3 +1,9 @@+"""
+Block-sparse runtime utilities for CUTE DSL kernels.
+
+This module contains runtime execution functions for block-sparse attention kernels.
+These utilities are used by CUTE DSL kernels to produce and consume block-sparse loads.
+"""
from typing import Callable, Optional
from functools... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/block_sparse_utils.py |
Generate docstrings with examples | # Adapted from https://github.com/mlcommons/training_results_v1.1/blob/main/NVIDIA/benchmarks/bert/implementations/pytorch/padding.py
import torch
import torch.nn.functional as F
from einops import rearrange, repeat
class IndexFirstAxis(torch.autograd.Function):
@staticmethod
def forward(ctx, input, indices)... | --- +++ @@ -96,6 +96,18 @@
def unpad_input(hidden_states, attention_mask, unused_mask=None):
+ """
+ Arguments:
+ hidden_states: (batch, seqlen, ...)
+ attention_mask: (batch, seqlen), bool / int, 1 means valid and 0 means not valid.
+ unused_mask: (batch, seqlen), bool / int, 1 means the... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/bert_padding.py |
Write beginner-friendly docstrings | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# A reimplementation of https://github.com/Dao-AILab/flash-attention/blob/main/hopper/mainloop_bwd_sm80.hpp
# from Cutlass C++ to Cute-DSL.
import math
from types import SimpleNamespace
from typing import Type, Callabl... | --- +++ @@ -47,6 +47,21 @@ AtomLayoutMdQ: int = 1,
V_in_regs: bool = False,
):
+ """Initializes the configuration for a flash attention v2 kernel.
+
+ All contiguous dimensions must be at least 16 bytes aligned which indicates the head dimension
+ should be a multiple of 8.
+
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_bwd.py |
Improve documentation using docstrings | # Manage Ahead-of-Time (AOT) compiled kernels
import fcntl
import hashlib
import logging
import os
import pickle
import sys
import tempfile
import time
from functools import lru_cache
from getpass import getuser
from pathlib import Path
from typing import Hashable, TypeAlias
import ctypes
import cutlass
import cutlas... | --- +++ @@ -57,6 +57,16 @@
@lru_cache(maxsize=1)
def _compute_source_fingerprint() -> str:
+ """
+ Hash all CuTe Python sources plus runtime ABI stamps into a short fingerprint.
+
+ The fingerprint changes whenever:
+ - Any .py file under flash_attn/cute is added, removed, renamed, or modified.
+ - Th... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/cache_utils.py |
Add documentation for all methods | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# A reimplementation of https://github.com/Dao-AILab/flash-attention/blob/main/hopper/flash_bwd_preprocess_kernel.h
# from Cutlass C++ to Cute-DSL.
#
# Computes D_i = (dO_i * O_i).sum(dim=-1), optionally adjusted for L... | --- +++ @@ -44,6 +44,17 @@ tile_m: int = 128,
num_threads: int = 256,
):
+ """
+ All contiguous dimensions must be at least 16 bytes aligned which indicates the head dimension
+ should be a multiple of 8.
+
+ :param head_dim: head dimension
+ :type head_dim: int
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_bwd_preprocess.py |
Annotate my code with docstrings | import argparse
import time
import torch
try:
import cudnn
except ImportError:
cudnn = None
from einops import rearrange
from flash_attn.cute.bench_utils import (
flops, attention_ref,
cudnn_fwd_setup, cudnn_bwd_setup,
)
try:
from flash_attn.flash_attn_interface import flash_attn_func, flash_att... | --- +++ @@ -41,6 +41,13 @@ # ── Autograd backward helper ────────────────────────────────────────────────
def _make_bwd_fn(fwd_fn, g, inputs):
+ """Run fwd once, return a closure that benchmarks backward.
+
+ Args:
+ fwd_fn: zero-arg callable that runs the forward pass (with autograd).
+ g: gradi... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/benchmarks/benchmark_attn.py |
Add docstrings to incomplete code | # Copyright (c) 2025, Ted Zadouri, Markus Hoehnerbach, Jay Shah, Tri Dao.
import math
from typing import Callable, Optional
from functools import partial
import cuda.bindings.driver as cuda
import cutlass
import cutlass.cute as cute
from cutlass.cute import FastDivmodDivisor
from cutlass import Float32, Int32, Int64,... | --- +++ @@ -2757,6 +2757,7 @@ aux_tensors=None,
fastdiv_mods=(None, None),
):
+ """Apply forward score modification for SM100 backward pass."""
# In bwd, S is computed as K @ Q.T so dimensions are (tile_n, tile_m)
cS = cute.make_identity_tensor((self.tile_n, self.tile_m))
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_bwd_sm100.py |
Add concise docstrings to each method | # Copyright (c) 2023, Tri Dao.
import torch
import torch.utils.benchmark as benchmark
def benchmark_forward(
fn, *inputs, repeats=10, desc="", verbose=True, amp=False, amp_dtype=torch.float16, **kwinputs
):
if verbose:
print(desc, "- Forward pass")
def amp_wrapper(*inputs, **kwinputs):
w... | --- +++ @@ -1,4 +1,5 @@ # Copyright (c) 2023, Tri Dao.
+"""Useful functions for writing test code."""
import torch
import torch.utils.benchmark as benchmark
@@ -7,6 +8,7 @@ def benchmark_forward(
fn, *inputs, repeats=10, desc="", verbose=True, amp=False, amp_dtype=torch.float16, **kwinputs
):
+ """Use Pyto... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/benchmark.py |
Create documentation for each function signature | # Copyright (c) 2025, Tri Dao
import math
from functools import partial
from typing import Optional, Tuple, Union
import torch
from torch import Tensor
from einops import rearrange, repeat
from flash_attn.ops.triton.rotary import apply_rotary
def rotate_half(x, interleaved=False):
if not interleaved:
x... | --- +++ @@ -21,6 +21,10 @@
def apply_rotary_emb_torch(x, cos, sin, interleaved=False):
+ """
+ x: (batch_size, seqlen, nheads, headdim)
+ cos, sin: (seqlen, rotary_dim / 2) or (batch_size, seqlen, rotary_dim / 2)
+ """
ro_dim = cos.shape[-1] * 2
assert ro_dim <= x.shape[-1]
cos = repeat(c... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/layers/rotary.py |
Write docstrings including parameters and return values | # Adapted from https://github.com/mlcommons/training_results_v1.1/blob/main/NVIDIA/benchmarks/bert/implementations/pytorch/fmha.py
import flash_attn_cuda
import torch
import torch.nn as nn
def convert_blockmask(blockmask, causal):
assert not causal
# TD [2022-05-13]: The indexing and sorting is very tricky
... | --- +++ @@ -5,6 +5,18 @@
def convert_blockmask(blockmask, causal):
+ """Convert from the 0-1 format to the format used by the CUDA code.
+ 0 means the block is skipped.
+ nonzero means the block is not skipped.
+ Argument:
+ blockmask: (row, col): a 0-1 tensor
+ Return:
+ blockmask_conv... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/flash_blocksparse_attn_interface.py |
Generate docstrings for each module | # Copyright (c) 2022, Tri Dao.
# This BERT implementation is based on our MLPerf 2.0 and MLPerf 2.1 BERT implementation.
# https://github.com/mlcommons/training_results_v2.0/blob/main/HazyResearch/benchmarks/bert/implementations/pytorch/modeling.py
# https://github.com/mlcommons/training_results_v2.1/blob/main/Azure-Ha... | --- +++ @@ -158,6 +158,10 @@ )
def forward(self, hidden_states, key_padding_mask=None, subset_mask=None):
+ """If subset_mask is not None, we only want output for the subset of the sequence.
+ This means that we only compute the last layer output for these tokens.
+ subset_mask: (bat... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/models/bert.py |
Add docstrings following best practices | # Copyright (c) 2025, Tri Dao.
from typing import Optional, Callable, TypeAlias
from dataclasses import dataclass
import cutlass
import cutlass.cute as cute
from cutlass import Float32, Int32, Uint32, const_expr
from quack import layout_utils
import flash_attn.cute.utils as utils
from flash_attn.cute.seqlen_info imp... | --- +++ @@ -17,12 +17,22 @@
@cute.jit
def r2p_bitmask_below(limit: Int32, s: int) -> Uint32:
+ """32-bit R2P bitmask keeping positions < limit (exclusive upper bound).
+
+ Positions 0..limit-1 in chunk `s` get bit=1 (keep), the rest bit=0 (mask).
+ Uses inline PTX to avoid shift-by-type-width UB.
+ """
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/mask.py |
Generate docstrings for exported functions | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# A reimplementation of https://github.com/Dao-AILab/flash-attention/blob/main/hopper/flash_fwd_combine_kernel.h
# from Cutlass C++ to Cute-DSL.
import math
from typing import Type, Optional
from functools import parti... | --- +++ @@ -30,6 +30,19 @@ num_threads: int = 256,
stages: int = 4,
):
+ """
+ Forward combine kernel for split attention computation.
+
+ :param dtype: output data type
+ :param dtype_partial: partial accumulation data type
+ :param head_dim: head dimension
+ ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_fwd_combine.py |
Write clean docstrings for readability | # Copyright (c) 2025, Tri Dao.
import cutlass
import cutlass.cute as cute
from quack import layout_utils
import flash_attn.cute.utils as utils
def pack_gqa_layout(T, qhead_per_kvhead, nheads_kv, head_idx):
head_stride = T.stride[head_idx]
shape_packed = (
(qhead_per_kvhead, T.shape[0]),
*[T... | --- +++ @@ -9,6 +9,17 @@
def pack_gqa_layout(T, qhead_per_kvhead, nheads_kv, head_idx):
+ """Reshape a tensor to fold qhead_per_kvhead into the seqlen dimension (mode 0).
+
+ The head dimension is at mode ``head_idx``. Modes before it (1..head_idx-1)
+ are kept as-is (e.g. headdim for Q/O tensors), and mo... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/pack_gqa.py |
Provide clean and structured docstrings | # Copyright (c) 2025, Tri Dao.
# import math
from typing import Optional
from dataclasses import dataclass
import cutlass.cute as cute
from cutlass import Boolean, Int32, const_expr
from cutlass.cutlass_dsl import if_generate, dsl_user_op
from cutlass.pipeline import PipelineState
from cutlass.pipeline import Pipelin... | --- +++ @@ -18,6 +18,11 @@
class PipelineStateSimple:
+ """
+ Pipeline state contains an index and phase bit corresponding to the current position in the circular buffer.
+ Use a single Int32 to store both the index and phase bit, then we use divmod to get the
+ index and phase. If stages is a power of ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/pipeline.py |
Add docstrings that explain purpose and usage |
import math
# H100 hardware limits
SMEM_LIMIT = 224 * 1024 # 228 KB minus ~3 KB for LSE, dPsum, mbarriers
REG_LIMITS = {2: 216, 3: 128} # per-WG budget: 2WG=240-24, 3WG=160-32
THREADS_PER_WG = 128
def _divisors(n):
return [d for d in range(1, n + 1) if n % d == 0]
def _acc_regs(M, N, num_wg):
return M *... | --- +++ @@ -1,3 +1,13 @@+"""Search feasible SM90 fwd/bwd attention configs for given (head_dim, head_dim_v).
+
+Enumerates tile sizes, swap modes, atom layouts, and staging options.
+Checks GMMA divisibility, register budget, and shared memory budget.
+
+Usage:
+ python flash_attn/cute/sm90_config_search.py --headdi... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/sm90_config_search.py |
Document functions with clear intent | import math
import re
from collections import OrderedDict
import torch
import torch.nn.functional as F
from transformers import GPT2Config, GPTBigCodeConfig, PretrainedConfig
def remap_state_dict_hf_bigcode(state_dict, config: PretrainedConfig):
# Word embedding and position embedding
def key_mapping_pos_em... | --- +++ @@ -8,6 +8,9 @@
def remap_state_dict_hf_bigcode(state_dict, config: PretrainedConfig):
+ """
+ Map the state_dict of a Huggingface BigCode model to be flash_attn compatible.
+ """
# Word embedding and position embedding
def key_mapping_pos_emb(key):
@@ -107,6 +110,11 @@
def inv_rema... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/models/bigcode.py |
Add docstrings for production code | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# [2025-07-04] Version in Cute-DSL, for Hopper and Blackwell. You'll need install nvidia-cutlass-dsl==4.2.0.
# Supported features:
# - BF16 & FP16 dtype
# - noncausal & causal attention
# - MHA, GQA, MQA
# - hdim 64, ... | --- +++ @@ -70,6 +70,7 @@ )
def _parse_arch_str(arch_str):
+ """Parse arch string (e.g. 'sm_80', 'sm_90a', '80', '100') to int (e.g. 80, 90, 100)."""
import re
match = re.match(r"^(?:sm_?|SM_?)?(\d+)(\d)([af]?)$", arch_str)
if not match:
@@ -80,6 +81,16 @@
@lru_cache(maxsize=None)
def _get_devic... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/interface.py |
Generate documentation strings for clarity | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# A reimplementation of
# https://github.com/Dao-AILab/flash-attention/blob/main/hopper/flash_fwd_kernel_sm80.h
# and https://github.com/Dao-AILab/flash-attention/blob/main/hopper/flash_fwd_kernel_sm90.h
# from Cutlass... | --- +++ @@ -56,6 +56,25 @@ has_aux_tensors: bool = False,
q_subtile_factor: int | None = None,
):
+ """Initializes the configuration for a flash attention kernel.
+
+ All contiguous dimensions must be at least 16 bytes aligned, which means that the head dimension
+ should be a... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_fwd.py |
Document all public functions with docstrings | # Copyright (c) 2025, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
# SM90 (Hopper) forward pass for flash attention, extracted from flash_fwd.py.
from types import SimpleNamespace
from typing import Callable, Optional
from functools import partial
import cuda.bindings.driver as cuda... | --- +++ @@ -166,6 +166,11 @@ # Always keep stream as the last parameter (EnvStream: obtained implicitly via TVM FFI).
stream: cuda.CUstream = None,
):
+ """Configures and launches the flash attention kernel.
+
+ mQ/mK/mV/mO has same data types(supports fp16 and bf16) and same layout:... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_fwd_sm90.py |
Add concise docstrings to each method | # Supported features:
# - BF16 & FP16 dtype
# - noncausal & causal attention
# - MHA, GQA, MQA
# - hdim 64, 96, 128, (192, 128).
# - varlen
# - sliding window
# - split-kv
# Unsupported features that will be added later:
# - page size != 128
# - more hdim (192, 256)
# Based on the cutlass example and cute-dsl example:
... | --- +++ @@ -241,6 +241,14 @@ self.buffer_align_bytes = 1024
def _setup_attributes(self):
+ """Set up configurations and parameters for the FMHA kernel operation.
+
+ This method initializes and configures various attributes required for the
+ execution of the fused multi-head attenti... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/flash_fwd_sm100.py |
Add docstrings with type hints explained | # Copyright (c) 2024, Tri Dao.
import logging
import math
import re
from collections import OrderedDict, namedtuple
from collections.abc import Sequence
from functools import partial
from typing import Dict, List
import torch
import torch.nn as nn
import torch.nn.functional as F
from einops import rearrange
from tran... | --- +++ @@ -309,6 +309,9 @@
class GPTPreTrainedModel(nn.Module):
+ """An abstract class to handle weights initialization and
+ a simple interface for dowloading and loading pretrained models.
+ """
def __init__(self, config, *inputs, **kwargs):
super().__init__()
@@ -335,6 +338,10 @@ ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/models/gpt.py |
Add docstrings following best practices | # Adapted from https://github.com/facebookresearch/xformers/blob/main/xformers/triton/k_activations.py
# Copyright (c) Facebook, Inc. and its affiliates. All rights reserved.
#
# This source code is licensed under the BSD license found in the
# LICENSE file in the root directory of this source tree.
import math
from e... | --- +++ @@ -70,6 +70,11 @@ # ReLU
@triton.jit
def relu(x):
+ """
+ ReLU_ activation function
+
+ .. _ReLU: https://pytorch.org/docs/stable/generated/torch.nn.ReLU.html
+ """
zero = 0.0
return tl.where(x >= 0, x, zero.to(x.dtype))
@@ -86,6 +91,11 @@
@triton.jit
def squared_relu(x):
+ """
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/triton/k_activations.py |
Document this module using docstrings | # Adapted from https://github.com/ELS-RD/kernl/blob/main/src/kernl/implementations/linear_layer.py
# and https://github.com/openai/triton/blob/master/python/triton/ops/matmul.py
from typing import Optional
import torch
import triton
import triton.language as tl
from triton.ops.matmul_perf_model import early_config_pru... | --- +++ @@ -165,6 +165,16 @@ ACTIVATION: tl.constexpr,
):
+ """
+ Kernel for computing Out = activation(A x W + C)
+ - Input has shape (M, K)
+ - Weight has shape (K, N)
+ - Bias has shape (N,)
+ - Output has shape (M, N)
+ - ActInputs (optional) has shape (M, N)
+ 'ActInputs' optionally ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/triton/linear.py |
Add detailed docstrings explaining each function | import math
import hydra
import torch
import torch.nn as nn
from einops import rearrange
from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input
from flash_attn.flash_blocksparse_attn_interface import (
convert_blockmask,
flash_blocksparse_attn_func,
)
class FlashBlocksparseAttention(nn... | --- +++ @@ -13,6 +13,15 @@
class FlashBlocksparseAttention(nn.Module):
+ """Implement the scaled dot product attention with softmax.
+ Arguments
+ ---------
+ softmax_temp: The temperature to use for the softmax attention.
+ (default: 1/sqrt(d_keys) where d_keys is computed at
+... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/flash_blocksparse_attention.py |
Add inline docstrings for readability | # Copyright (c) 2024, Tri Dao.
import torch
import torch.nn as nn
from flash_attn.ops.triton.cross_entropy import cross_entropy_loss
class CrossEntropyLoss(nn.Module):
def __init__(
self,
ignore_index=-100,
reduction="mean",
label_smoothing=0.0,
logit_scale=1.0,
l... | --- +++ @@ -18,6 +18,20 @@ process_group=None,
return_z_loss=False,
):
+ """
+ Arguments:
+ ignore_index: int. If labels == ignore_index, the loss is set to 0.0.
+ label_smoothing: float
+ lse_square_scale: float. If > 0, we add lse_square_scale * lse... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/losses/cross_entropy.py |
Can you add docstrings to this Python file? | # Copyright (c) 2025, Tri Dao.
# Ported Cutlass code from C++ to Python:
# https://github.com/NVIDIA/cutlass/blob/main/include/cute/arch/mma_sm100_desc.hpp
# https://github.com/NVIDIA/cutlass/blob/main/include/cute/atom/mma_traits_sm100.hpp
from enum import IntEnum
import cutlass
import cutlass.cute as cute
# ------... | --- +++ @@ -66,6 +66,9 @@
def to_UMMA_format(cutlass_type) -> int:
+ """
+ Map a CUTLASS scalar class to the 3-bit encoding for Matrix A/B.
+ """
if cutlass_type is cutlass.Int8:
return S8Format.INT8
# Unsigned 8-bit (if available in your CUTLASS build)
@@ -88,6 +91,9 @@
def to_C_for... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/mma_sm100_desc.py |
Add standardized docstrings across the file | from typing import Optional
from dataclasses import dataclass
import cutlass
import cutlass.cute as cute
from cutlass import Int32, const_expr
from quack import copy_utils
"""
This consolidates all the info related to sequence length. This is so that we can do all
the gmem reads once at the beginning of each tile, r... | --- +++ @@ -52,6 +52,7 @@ padded: cutlass.Constexpr[bool] = False,
multiple: int = 1,
) -> cute.Tensor:
+ """Offset a tensor by batch index. batch dim is at position `dim`, seqlen is at dim=0."""
if const_expr(not self.has_cu_seqlens):
idx = (None,) * dim + (batch_idx,... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/seqlen_info.py |
Generate NumPy-style docstrings |
import math
import torch
import triton
import triton.language as tl
# Disabling autotune for now, set num_warps=4 if headdim=64 and num_warps=8 if headdim=128
# @triton.autotune(
# configs=[
# triton.Config({"BLOCK_M": 128, "BLOCK_N": 128}, num_warps=4, num_stages=1),
# # This config has a race ... | --- +++ @@ -1,3 +1,43 @@+"""
+*Experimental* implementation of FlashAttention in Triton.
+Tested with triton==2.0.0.dev20221202.
+Triton 2.0 has a new backend (MLIR) but seems like it doesn't yet work for head dimensions
+other than 64:
+https://github.com/openai/triton/blob/d376020f90002757eea3ea9475d4f7cfc2ec5ead/pyt... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/flash_attn_triton.py |
Write Python docstrings for this snippet | # Copyright (c) 2023, Tri Dao.
# Adapted from https://github.com/NVIDIA/Megatron-LM/blob/0bb597b42c53355a567aba2a1357cc34b9d99ddd/megatron/text_generation/forward_step.py#L31
import gc
import time
from collections import namedtuple
from dataclasses import dataclass, field
from functools import partial
from typing impor... | --- +++ @@ -22,6 +22,8 @@
@dataclass
class InferenceParams:
+ """Inference parameters that are passed to the main model in order
+ to efficienly calculate and store the context during inference."""
max_seqlen: int
max_batch_size: int
@@ -41,6 +43,7 @@ # https://github.com/NVIDIA/Megatron-LM/blob/0b... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/utils/generation.py |
Add concise docstrings to each method | # Copyright (c) 2023, Tri Dao.
from typing import Optional, Sequence, Tuple, Union
import torch
import torch.nn as nn
import os
# isort: off
# We need to import the CUDA kernels after importing torch
USE_TRITON_ROCM = os.getenv("FLASH_ATTENTION_TRITON_AMD_ENABLE", "FALSE") == "TRUE"
if USE_TRITON_ROCM:
from aite... | --- +++ @@ -1016,6 +1016,40 @@ deterministic=False,
return_attn_probs=False,
):
+ """dropout_p should be set to 0.0 during evaluation
+ If Q, K, V are already stacked into 1 tensor, this function will be faster than
+ calling flash_attn_func on Q, K, V since the backward pass avoids explicit concaten... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/flash_attn_interface.py |
Add detailed documentation for each class | # Copyright (c) 2024, Tri Dao.
from functools import partial
from typing import Optional
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch import Tensor
from torchvision.ops import StochasticDepth
from flash_attn.modules.mha import MHA
from flash_attn.modules.mlp import Mlp
try:
from... | --- +++ @@ -37,6 +37,23 @@ sequence_parallel=False,
mark_shared_params=False,
):
+ """
+ For prenorm=True, this Block has a slightly different structure compared to a regular
+ prenorm Transformer block.
+ The standard block is: LN -> MHA -> Dropout -> Add -> LN -> MLP ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/modules/block.py |
Add standardized docstrings across the file | # Copyright (c) 2023, Tri Dao.
import json
import math
import os
import re
from collections import OrderedDict
from pathlib import Path
from typing import Dict, List, Union
import torch
import torch.nn.functional as F
from sentencepiece import SentencePieceProcessor
from transformers import GPT2Config, LlamaConfig
f... | --- +++ @@ -19,6 +19,10 @@ def remap_state_dict_meta_llama(
state_dict: Dict[str, torch.Tensor], config: GPT2Config
) -> Dict[str, torch.Tensor]:
+ """Convert the state_dict in Meta format to standard GPT format.
+
+ This function modifies state_dict in place.
+ """
def key_mapping_layers(key):
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/models/llama.py |
Generate NumPy-style docstrings | # Copyright (c) 2022, Tri Dao.
# Inspired by / adapted from https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py
import math
import re
from collections import OrderedDict
from copy import deepcopy
from functools import partial
import torch
import torch.nn as nn
import torch.n... | --- +++ @@ -95,6 +95,10 @@
class VisionTransformer(nn.Module):
+ """Vision Transformer
+ A PyTorch impl of : `An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale`
+ - https://arxiv.org/abs/2010.11929
+ """
def __init__(
self,
@@ -125,6 +129,29 @@ fused_... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/models/vit.py |
Write documentation strings for class attributes | # Copyright (c) 2025, Tri Dao.
import math
import operator
from typing import Tuple
from dataclasses import dataclass
import cutlass
import cutlass.cute as cute
from cutlass import Float32
from quack import layout_utils
import flash_attn.cute.utils as utils
from quack.cute_dsl_utils import ParamsBase
from flash_attn... | --- +++ @@ -56,6 +56,13 @@ is_first: cutlass.Constexpr[bool] = False,
check_inf: cutlass.Constexpr[bool] = True,
) -> cute.Tensor:
+ """Apply online softmax and return the row_scale to rescale O.
+
+ :param acc_S: acc_S tensor
+ :type acc_S: cute.Tensor
+ :param is_firs... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/softmax.py |
Generate docstrings for each module | # Copyright (c) 2022, Tri Dao.
import torch
import torch.nn as nn
from einops import rearrange
from torch import Tensor
from flash_attn.utils.distributed import all_reduce, reduce_scatter
class GPT2Embeddings(nn.Module):
def __init__(
self,
embed_dim,
vocab_size,
max_position_emb... | --- +++ @@ -19,6 +19,11 @@ device=None,
dtype=None,
):
+ """
+ If max_position_embeddings <= 0, there's no position embeddings
+ If word_embe_proj_dim is not None (e.g., OPT-350m), we embed to that dimension
+ the project up to embed_dim
+ """
factor... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/modules/embedding.py |
Generate docstrings for each module | import subprocess
from pathlib import Path
from typing import List
import matplotlib.pyplot as plt
import seaborn as sn
import torch
import wandb
from pytorch_lightning import Callback, Trainer
from pytorch_lightning.loggers import LoggerCollection, WandbLogger
from pytorch_lightning.utilities import rank_zero_only
fr... | --- +++ @@ -14,6 +14,7 @@
def get_wandb_logger(trainer: Trainer) -> WandbLogger:
+ """Safely get Weights&Biases logger from Trainer."""
if trainer.fast_dev_run:
raise Exception(
@@ -34,6 +35,7 @@
class WatchModel(Callback):
+ """Make wandb watch model at the beginning of the run."""
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/callbacks/wandb_callbacks.py |
Write clean docstrings for readability | # Copyright (c) 2023, Tri Dao.
import math
from functools import partial
import torch
import torch.nn as nn
from einops import rearrange, repeat
from flash_attn.utils.distributed import get_dim_for_local_rank
try:
from flash_attn import (
flash_attn_kvpacked_func,
flash_attn_qkvpacked_func,
... | --- +++ @@ -51,6 +51,15 @@
class FlashSelfAttention(nn.Module):
+ """Implement the scaled dot product attention with softmax.
+ Arguments
+ ---------
+ softmax_scale: The temperature to use for the softmax attention.
+ (default: 1/sqrt(d_keys) where d_keys is computed at
+ ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/modules/mha.py |
Generate docstrings with parameter types | # Copyright (c) 2025, Tri Dao.
import math
import hashlib
import inspect
from typing import Type, Callable, Optional, Tuple, overload
import cutlass
import cutlass.cute as cute
from cutlass import Float32, const_expr
from cutlass.cute import FastDivmodDivisor
from cutlass.cutlass_dsl import T, dsl_user_op
from cutla... | --- +++ @@ -57,6 +57,7 @@
def _compute_base_hash(func: Callable) -> str:
+ """Compute hash from source code or bytecode and closure values."""
try:
data = inspect.getsource(func).encode()
except (OSError, TypeError):
@@ -77,6 +78,12 @@ def hash_callable(
func: Callable, mixer_attrs: Tuple... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/cute/utils.py |
Write Python docstrings for this snippet | # The triton fused matmul + sqrelu is faster for fp16 but slower for bf16, compared
# to naive implementation.
import fused_dense_lib as fused_dense_cuda
import torch
import torch.nn as nn
import torch.nn.functional as F
from flash_attn.utils.torch import custom_fwd, custom_bwd
from flash_attn.ops.activations import s... | --- +++ @@ -14,6 +14,11 @@ @staticmethod
@custom_fwd
def forward(ctx, x, weight1, bias1, weight2, bias2, checkpoint_lvl=0):
+ """checkpoint_lvl:
+ 0: no recomputation in the bwd
+ 1: recompute gelu_out in the bwd
+ 2: recompute act_input and gelu_out in the bwd
+ """
... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/triton/mlp.py |
Help me add docstrings to my project | # Copied from https://github.com/stanford-crfm/mistral/blob/main/src/corpora/detokenization.py
# Which was originally from https://github.com/NVIDIA/Megatron-LM/blob/aed2f75e209e525c842aec7c044af7acae2a4614/tasks/zeroshot_gpt/detokenizer.py
import re
def wikitext_detokenize(string: str) -> str:
# Contractions
... | --- +++ @@ -1,10 +1,17 @@ # Copied from https://github.com/stanford-crfm/mistral/blob/main/src/corpora/detokenization.py
# Which was originally from https://github.com/NVIDIA/Megatron-LM/blob/aed2f75e209e525c842aec7c044af7acae2a4614/tasks/zeroshot_gpt/detokenizer.py
+"""
+Handle detokenization for different dataset ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/datamodules/datasets/detokenizer.py |
Add concise docstrings to each method | # Adapted from https://github.com/PyTorchLightning/lightning-bolts/blob/master/pl_bolts/datamodules/imagenet_datamodule.py
import os
from pathlib import Path
from typing import Any, List, Union, Callable, Optional
import torch
from torch.utils.data import Dataset, DataLoader, SequentialSampler
from torch.utils.data.da... | --- +++ @@ -17,6 +17,10 @@ class DictDataset(Dataset):
def __init__(self, dataset_dict, length=None):
+ """dataset_dict: dictionary mapping from index to batch
+ length is used in the case of DistributedSampler: e.g. the dataset could have size 1k, but
+ with 8 GPUs the dataset_dict would on... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/datamodules/imagenet.py |
Create documentation strings for testing functions | from typing import Optional
import torch
from torch import Tensor
from torch.distributed import ProcessGroup
# `all_gather_into_tensor` and `reduce_scatter_tensor` are new placeholders for
# `_all_gather_base` and `_reduce_scatter_base`. They require the most recent
# version of PyTorch. The following 4 lines are for... | --- +++ @@ -47,6 +47,7 @@
class AllGatherFunc(torch.autograd.Function):
+ """Gather the input from sequence parallel region and concatenate."""
@staticmethod
def forward(ctx, input_: Tensor, process_group: ProcessGroup) -> Tensor:
@@ -65,6 +66,7 @@
class ReduceScatterFunc(torch.autograd.Function)... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/utils/distributed.py |
Generate docstrings for script automation | # Copyright (c) 2022, Tri Dao.
# Adapted from https://github.com/NVIDIA/apex/blob/master/apex/contrib/layer_norm/layer_norm.py
import torch
from torch.nn import init
from flash_attn.ops.layer_norm import (
DropoutAddLayerNormFn,
DropoutAddLayerNormParallelResidualFn,
DropoutAddLayerNormSubsetFn,
)
def r... | --- +++ @@ -30,6 +30,9 @@ residual_in_fp32=False,
return_dropout_mask=False,
):
+ """residual_in_fp32 only has an effect if residual is None.
+ Otherwise residual dtype is residual.dtype.
+ """
return DropoutAddLayerNormFn.apply(
x0,
residual,
@@ -62,6 +65,9 @@ residual_in_... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/rms_norm.py |
Write beginner-friendly docstrings | # Copyright (c) 2022, Tri Dao.
# Adapted from https://github.com/NVIDIA/apex/blob/master/apex/contrib/layer_norm/layer_norm.py
import dropout_layer_norm
import torch
from torch.nn import init
def maybe_align(x, alignment_in_bytes=16):
# TD [2023-07-04] I'm not 100% sure that clone will align the memory
# htt... | --- +++ @@ -7,6 +7,7 @@
def maybe_align(x, alignment_in_bytes=16):
+ """Assume that x already has last dim divisible by alignment_in_bytes"""
# TD [2023-07-04] I'm not 100% sure that clone will align the memory
# https://discuss.pytorch.org/t/how-to-ensure-that-tensor-data-ptr-is-aligned-to-16-bytes/18... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/layer_norm.py |
Add docstrings that explain purpose and usage | from typing import Any, Dict, Optional
import torch
from torch import Tensor
from torchmetrics import Metric
class NumTokens(Metric):
# TODO: how do we prevent the reset between the epochs? The reset happens on the 1st batch
# of the next epoch.
# Right now the hack is that we override reset(), which wo... | --- +++ @@ -7,6 +7,8 @@
class NumTokens(Metric):
+ """Keep track of how many tokens we've seen.
+ """
# TODO: how do we prevent the reset between the epochs? The reset happens on the 1st batch
# of the next epoch.
# Right now the hack is that we override reset(), which would mess up the forward... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/metrics/num_tokens.py |
Add documentation for all methods | # Copyright (c) 2023, Tri Dao.
# Inspired by https://github.com/NVIDIA/apex/blob/master/apex/fused_dense/fused_dense.py
# We make it work with pytorch amp and with bfloat16.
# The TensorParallel linear modules are inspired by https://github.com/NVIDIA/apex/blob/master/apex/transformer/tensor_parallel/layers.py
from fun... | --- +++ @@ -30,6 +30,10 @@ def forward(
ctx, x, weight, bias, return_residual=False, process_group=None, sequence_parallel=True
):
+ """
+ If process_group is not None and sequence_parallel=True, we're doing Tensor Parallel
+ with sequence parallelism: we do an all_gather_raw of x... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/flash_attn/ops/fused_dense.py |
Add concise docstrings to each method | # Inspired by https://github.com/NVIDIA/NeMo/blob/main/nemo/collections/common/metrics/perplexity.py
# But we compute the perplexity correctly: exp(average(nll)), not average(exp(nll))
# Also adapted from https://github.com/Lightning-AI/metrics/blob/master/src/torchmetrics/text/perplexity.py
# But we pass in the loss t... | --- +++ @@ -19,6 +19,21 @@
class Perplexity(Metric):
+ r"""
+ Perplexity measures how well a language model predicts a text sample. It's calculated as the average number of bits
+ per word a model needs to represent the sample.
+ Args:
+ kwargs:
+ Additional keyword arguments, see :ref... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/metrics/perplexity.py |
Add docstrings to make code maintainable | # Copyright (c) 2024, Jay Shah, Ganesh Bikshandi, Ying Zhang, Vijay Thakkar, Pradeep Ramani, Tri Dao.
import sys
import warnings
import os
import stat
import re
import shutil
import ast
from pathlib import Path
from packaging.version import parse, Version
import platform
import sysconfig
import tarfile
import itertool... | --- +++ @@ -135,6 +135,20 @@ with_cuda,
**kwargs, # kwargs (ignored) to absorb new flags in torch.utils.cpp_extension
) -> None:
+ r"""Write a ninja file that does the desired compiling and linking.
+
+ `path`: Where to write this file
+ `cfl... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/hopper/setup.py |
Create simple docstrings for beginners | # Copyright (c) 2023, Tri Dao.
from typing import Optional, Union, List, Tuple
import os
import torch
import torch.nn as nn
USE_TRITON_ROCM = os.getenv("FLASH_ATTENTION_TRITON_AMD_ENABLE", "FALSE") == "TRUE"
if USE_TRITON_ROCM:
from aiter.ops.triton._triton_kernels.flash_attn_triton_amd import flash_attn_3 as f... | --- +++ @@ -179,6 +179,10 @@ pack_gqa: Optional[bool] = None,
sm_margin: int = 0,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+ """
+ Symbolic fake implementation of flash attention forward.
+ Returns tensors with the correct shapes and dtypes without actual computation.
+ ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/hopper/flash_attn_interface.py |
Create docstrings for API functions | # Copied from https://github.com/fadel/pytorch_ema/blob/master/torch_ema/ema.py
from __future__ import division
from __future__ import unicode_literals
from typing import Iterable, Optional
import weakref
import copy
import contextlib
import torch
def to_float_maybe(x):
return x.float() if x.dtype in [torch.flo... | --- +++ @@ -17,6 +17,15 @@ # Partially based on:
# https://github.com/tensorflow/tensorflow/blob/r1.13/tensorflow/python/training/moving_averages.py
class ExponentialMovingAverage:
+ """
+ Maintains (exponential) moving average of a set of parameters.
+ Args:
+ parameters: Iterable of `torch.nn.Parame... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/utils/ema.py |
Help me add docstrings to my project | # Adapted from https://github.com/mlcommons/training_results_v1.1/blob/main/NVIDIA/benchmarks/bert/implementations/pytorch/padding.py
import torch
import torch.nn.functional as F
from einops import rearrange
def unpad_input(hidden_states, attention_mask, unused_mask=None):
all_masks = (attention_mask + unused_ma... | --- +++ @@ -6,6 +6,18 @@
def unpad_input(hidden_states, attention_mask, unused_mask=None):
+ """
+ Arguments:
+ hidden_states: (batch, seqlen, ...)
+ attention_mask: (batch, seqlen), bool / int, 1 means valid and 0 means not valid.
+ unused_mask: (batch, seqlen), bool / int, 1 means the e... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/hopper/padding.py |
Document this code for team use | import logging
import warnings
from typing import List, Sequence
import pytorch_lightning as pl
import rich.syntax
import rich.tree
from omegaconf import DictConfig, OmegaConf
from pytorch_lightning.utilities import rank_zero_only
# Copied from https://docs.python.org/3/howto/logging-cookbook.html#using-a-context-ma... | --- +++ @@ -35,6 +35,7 @@
def get_logger(name=__name__) -> logging.Logger:
+ """Initializes multi-GPU-friendly python logger."""
logger = logging.getLogger(name)
@@ -47,6 +48,14 @@
def extras(config: DictConfig) -> None:
+ """A couple of optional utilities, controlled by main config file:
+ - ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/utils/utils.py |
Add docstrings for internal functions | # Meant to work with Apex's DistributeFusedAdam
from typing import Any, Callable, Dict, List, Optional, Union
from pathlib import Path
import types
import torch
from torch.optim.optimizer import Optimizer
from torch.optim import LBFGS
from apex.contrib.optimizers.distributed_fused_adam import DistributedFusedAdam
f... | --- +++ @@ -62,6 +62,7 @@ return closure_result
def clip_grad_by_norm(self, optimizer: DistributedFusedAdam, clip_val: Union[int, float]) -> None:
+ """Clip gradients by norm."""
# DistributedFusedAdam wants list, not generator
# Gradients have not be scaled, so we need to scale ... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/utils/ddp_zero2.py |
Generate docstrings for this script | # Meant to work with Pytorch's ZeroRedundancyOptimizer
from typing import Any, Callable, Dict, List, Optional, Union
from pathlib import Path
import torch
from torch.optim.optimizer import Optimizer
from torch.distributed.optim import ZeroRedundancyOptimizer
from pytorch_lightning.strategies.ddp import DDPStrategy
f... | --- +++ @@ -60,6 +60,9 @@
class DDPStrategyZero1(DDPStrategy):
+ """To use ZeroRedundancyOptimizer, we need to shard the optimizer states when
+ saving/loading checkpoints.
+ """
strategy_name = "ddp_zero1"
@@ -74,6 +77,12 @@ def save_checkpoint(
self, checkpoint: Dict[str, Any], file... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/utils/ddp_zero1.py |
Fully document this Python code with docstrings | # Copyright (c) 2023, Tri Dao.
import sys
import functools
import warnings
import os
import re
import ast
import glob
import shutil
from pathlib import Path
from typing import Literal, Optional
from packaging.version import parse, Version
import platform
from setuptools import setup, find_packages
import subprocess
... | --- +++ @@ -74,6 +74,9 @@
def get_platform():
+ """
+ Returns the platform name as used in wheel filenames.
+ """
if sys.platform.startswith("linux"):
return f'linux_{platform.uname().machine}'
elif sys.platform == "darwin":
@@ -95,6 +98,14 @@
def add_cuda_gencodes(cc_flag, archs, ba... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/setup.py |
Generate docstrings with parameter types | # Copied from https://github.com/NVIDIA/DeepLearningExamples/blob/master/PyTorch/LanguageModeling/Transformer-XL/pytorch/utils/distributed.py
# Copyright (c) 2019-2020, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in com... | --- +++ @@ -21,6 +21,11 @@
def init_distributed(cuda):
+ """
+ Initializes distributed backend.
+ :param cuda: (bool) if True initializes nccl backend, if False initializes
+ gloo backend
+ """
world_size = int(os.environ.get('WORLD_SIZE', 1))
distributed = (world_size > 1)
if dist... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/utils/distributed.py |
Write docstrings describing functionality | # Inspired by https://github.com/PyTorchLightning/pytorch-lightning/blob/master/pytorch_lightning/callbacks/stochastic_weight_avg.py
# https://github.com/PyTorchLightning/Lightning-Bolts/blob/master/pl_bolts/callbacks/byol_updates.py
# https://forums.pytorchlightning.ai/t/adopting-exponential-moving-average-ema-for-pl-... | --- +++ @@ -14,7 +14,14 @@
class EMACallback(Callback):
+ """TD [2021-08-31]: saving and loading from checkpoint should work.
+ """
def __init__(self, decay: float, use_num_updates: bool = True):
+ """
+ decay: The exponential decay.
+ use_num_updates: Whether to use number of update... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/callbacks/ema.py |
Provide clean and structured docstrings | import math
from typing import List, Tuple
import torch
import torchvision.transforms as T
from PIL import Image, ImageOps
from transformers import AutoProcessor, BatchFeature, LlamaTokenizerFast
from transformers.processing_utils import ProcessorMixin
from config import IMAGE_SIZE, BASE_SIZE, CROP_MODE, MIN_CROPS, MA... | --- +++ @@ -245,6 +245,24 @@ inference_mode: bool = True,
**kwargs,
):
+ """
+
+ Args:
+ prompt (str): the formatted prompt;
+ conversations (List[Dict]): conversations with a list of messages;
+ images (List[ImageType]): the list of images;
+ ... | https://raw.githubusercontent.com/deepseek-ai/DeepSeek-OCR/HEAD/DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py |
Write docstrings that follow conventions | from typing import List, Optional, Sequence
from pathlib import Path
import hydra
from omegaconf import OmegaConf, DictConfig
from pytorch_lightning import (
Callback,
LightningDataModule,
LightningModule,
Trainer,
seed_everything,
)
from pytorch_lightning.loggers import LightningLoggerBase
from s... | --- +++ @@ -18,6 +18,8 @@
def last_modification_time(path):
+ """Including files / directory 1-level below the path
+ """
path = Path(path)
if path.is_file():
return path.stat().st_mtime
@@ -28,6 +30,15 @@
def train(config: DictConfig) -> Optional[float]:
+ """Contains training pipel... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/train.py |
Add missing documentation to my Python functions | # Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
import torch
import torch.nn as nn
import torch.nn.functional as F
from typing import Optional, Tuple, Type
from functool... | --- +++ @@ -94,6 +94,24 @@ window_size: int = 0,
global_attn_indexes: Tuple[int, ...] = (),
) -> None:
+ """
+ Args:
+ img_size (int): Input image size.
+ patch_size (int): Patch size.
+ in_chans (int): Number of input image channels.
+ emb... | https://raw.githubusercontent.com/deepseek-ai/DeepSeek-OCR/HEAD/DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py |
Generate docstrings for this script | import math
from functools import partial
from collections import namedtuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn.modules.utils import _pair
import hydra
from einops import reduce, rearrange
def pooling(x, pooling_mode='CLS', key_padding_mask=None, batch_first=True):
... | --- +++ @@ -47,6 +47,7 @@
class ClassificationHeadLinear(nn.Module):
+ """Head for sentence-level classification tasks."""
def __init__(self, d_model, num_classes, pooling_mode='MEAN',
batch_first=False, **kwargs):
@@ -57,6 +58,9 @@ self.out_proj = nn.Linear(d_model, num_classes)... | https://raw.githubusercontent.com/Dao-AILab/flash-attention/HEAD/training/src/models/modules/seq_common.py |
Add well-formatted docstrings | # No unicode_literals
import json
import os
import unicodedata
from binascii import hexlify, unhexlify
from cryptography.hazmat.primitives.kdf import hkdf
from cryptography.hazmat.primitives import hashes
from attr import attrs, attrib
def HKDF(skm, outlen, salt=None, CTXinfo=b""):
return hkdf.HKDF(
hashe... | --- +++ @@ -9,6 +9,18 @@
def HKDF(skm, outlen, salt=None, CTXinfo=b""):
+ """
+ Return the RFC5869 'HMAC-based Key Derivation Function' result of
+ using the given `salt`, tag from `CTXinfo` and secret from `skm`
+ with a SHA256 hash.
+
+ :param bytes skm: the input key material
+ :param int outle... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/util.py |
Create simple docstrings for beginners | from attrs import frozen, Factory
@frozen
class Disconnected:
pass
@frozen
class Connecting:
url: str
last_attempt: int # unix-timestamp when we last tried connecting
@frozen
class Connected:
url: str
@frozen
class Failed:
reason: str
@frozen
class Closed:
@frozen
class NoPeer:
@froze... | --- +++ @@ -24,23 +24,39 @@
@frozen
class Closed:
+ """
+ We have purposely shut down our connection to the server
+ """
@frozen
class NoPeer:
+ """
+ We have yet to see a peer
+ """
@frozen
class StoppedPeer:
+ """
+ We have disconnected from the peer (probably on purpose)
+ "... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/_status.py |
Write reusable docstrings | from collections import deque, defaultdict
from attr import attrs, attrib
from attr.validators import instance_of
from zope.interface import implementer
from twisted.internet.defer import inlineCallbacks
from twisted.internet.interfaces import (ITransport, IProducer, IConsumer,
... | --- +++ @@ -51,6 +51,10 @@
class UnexpectedSubprotocol(Exception):
+ """
+ The peer sends an OPEN for a subprotocol name application code
+ indicates will never exist
+ """
@implementer(IAddress)
@@ -61,6 +65,13 @@ @implementer(IAddress)
@attrs
class SubchannelAddress:
+ """
+ All subchanne... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/_dilation/subchannel.py |
Fully document this Python code with docstrings | # Version: 0.29
# pylint:disable=invalid-name,import-outside-toplevel,missing-function-docstring
# pylint:disable=missing-class-docstring,too-many-branches,too-many-statements
# pylint:disable=raise-missing-from,too-many-lines,too-many-locals,import-error
# pylint:disable=too-few-public-methods,redefined-outer-name,co... | --- +++ @@ -1,5 +1,306 @@ # Version: 0.29
+"""The Versioneer - like a rocketeer, but for versions.
+
+The Versioneer
+==============
+
+* like a rocketeer, but for versions!
+* https://github.com/python-versioneer/python-versioneer
+* Brian Warner
+* License: Public Domain (Unlicense)
+* Compatible with: Python 3.7, ... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/versioneer.py |
Generate docstrings for this script | import os
from collections import deque
from attr import attrs, attrib, evolve, define, field
from attr.validators import instance_of, optional
from automat import MethodicalMachine
from zope.interface import implementer
from twisted.internet.defer import Deferred, inlineCallbacks
from twisted.python import log, failur... | --- +++ @@ -55,6 +55,15 @@
@define
class DilatedWormhole:
+ """
+ Represents actions available once a wormhole has been successfully dilated.
+
+ New subchannels to the other peer may be established by first
+ obtaining an `IStreamClientEndpoint` from the
+ `subprotocol_connector_for("subproto-name")`... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/_dilation/manager.py |
Document helper functions with docstrings | from zope.interface import Interface
# These interfaces are private: we use them as markers to detect
# swapped argument bugs in the various .wire() calls
class IWormhole(Interface):
def got_welcome(welcome):
pass
def got_code(code):
pass
def got_key(key):
pass
def got_ver... | --- +++ @@ -5,6 +5,7 @@
class IWormhole(Interface):
+ """Internal: this contains the methods invoked 'from below'."""
def got_welcome(welcome):
pass
@@ -29,20 +30,84 @@
class IWormholeDelegate(Interface):
+ """
+ Must be implemented by the object passed as "delegate=" to wormhole.create... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/_interfaces.py |
Document my Python code with docstrings | # This file helps to compute a version number in source trees obtained from
# git-archive tarball (such as those provided by githubs download-from-tag
# feature). Distribution tarballs (built by setup.py sdist) and build
# directories (produced by setup.py build) will contain a much shorter file
# that just contains th... | --- +++ @@ -8,6 +8,7 @@ # Generated by versioneer-0.29
# https://github.com/python-versioneer/python-versioneer
+"""Git implementation of _version.py."""
import errno
import os
@@ -19,6 +20,7 @@
def get_keywords() -> dict[str, str]:
+ """Get the keywords needed to look up the version information."""
... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/_version.py |
Generate docstrings with examples | import os
import sys
from attr import attrib, attrs
from twisted.python import failure
from twisted.internet.task import Cooperator
from zope.interface import implementer
from ._boss import Boss
from ._dilation.manager import DILATION_VERSIONS
from ._dilation.connector import Connector
from ._interfaces import IDefer... | --- +++ @@ -76,6 +76,12 @@ self._boss.send(plaintext)
def derive_key(self, purpose, length):
+ """Derive a new key from the established wormhole channel for some
+ other purpose. This is a deterministic randomized function of the
+ session key and the 'purpose' string (unicode/py3-st... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/wormhole.py |
Write docstrings for algorithm functions | import os
import time
start = time.time()
from sys import stderr, stdout # noqa: E402
from textwrap import dedent, fill # noqa: E402
import click # noqa: E402
from twisted.internet.defer import inlineCallbacks, maybeDeferred # noqa: E402
from twisted.internet.task import react # noqa: E402
from twisted.python.fa... | --- +++ @@ -22,6 +22,9 @@
class Config:
+ """
+ Union of config options that we pass down to (sub) commands.
+ """
def __init__(self):
# This only holds attributes which are *not* set by CLI arguments.
@@ -113,6 +116,13 @@ )
@click.pass_context
def wormhole(context, dump_timing, transit_h... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/cli/cli.py |
Create documentation strings for testing functions | import hashlib
import os
import sys
import tempfile
import zipfile
from humanize import naturalsize
from tqdm import tqdm
from twisted.internet import reactor
from twisted.internet.defer import inlineCallbacks
from twisted.python import log
from wormhole import __version__, create, input_with_completion
from ..errors... | --- +++ @@ -34,6 +34,13 @@
def receive(args, reactor=reactor, _debug_stash_wormhole=None):
+ """I implement 'wormhole receive'. I return a Deferred that fires with
+ None (for success), or signals one of the following errors:
+ * WrongPasswordError: the two sides didn't use matching passwords
+ * Timeou... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/cli/cmd_receive.py |
Add docstrings for better understanding | class WormholeError(Exception):
class UnsendableFileError(Exception):
class ServerError(WormholeError):
class ServerConnectionError(WormholeError):
def __init__(self, url, reason):
self.url = url
self.reason = reason
def __str__(self):
return str(self.reason)
class Timeout(Worm... | --- +++ @@ -1,13 +1,24 @@ class WormholeError(Exception):
+ """Parent class for all wormhole-related errors"""
class UnsendableFileError(Exception):
+ """
+ A file you wanted to send couldn't be read, maybe because it's not
+ a file, or because it's a symlink that points to something
+ that doesn't ... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/errors.py |
Generate documentation strings for clarity | import os
import socket
import sys
import time
from binascii import hexlify, unhexlify
from collections import deque
from nacl.secret import SecretBox
from twisted.internet import (address, defer, endpoints, error, interfaces,
protocol, task)
from twisted.internet.defer import inlineCallb... | --- +++ @@ -330,6 +330,18 @@ # Helper methods
def connectConsumer(self, consumer, expected=None):
+ """Helper method to glue an instance of e.g. t.p.ftp.FileConsumer to
+ us. Inbound records will be written as bytes to the consumer.
+
+ Set 'expected' to an integer to automatically disco... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/transit.py |
Add docstrings following best practices | import json
from twisted.internet.defer import inlineCallbacks
from . import wormhole
from .tor_manager import get_tor
@inlineCallbacks
def receive(reactor,
appid,
relay_url,
code,
use_tor=False,
launch_tor=False,
tor_control_port=None,
... | --- +++ @@ -15,6 +15,26 @@ launch_tor=False,
tor_control_port=None,
on_code=None):
+ """
+ This is a convenience API which returns a Deferred that callbacks
+ with a single chunk of data from another wormhole (and then closes
+ the wormhole). Under the hood, it's just u... | https://raw.githubusercontent.com/magic-wormhole/magic-wormhole/HEAD/src/wormhole/xfer_util.py |
Add return value explanations in docstrings | #
# Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,17 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""A non-blocking, single-threaded HTTP server.
+
+Typical applications have little direct interaction with the `HTTPServer`
+class except to start a server at the beginning of the process
+(and... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/httpserver.py |
Add docstrings to clarify complex logic | # Copyright 2015 The Tornado Authors
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in w... | --- +++ @@ -12,6 +12,18 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""Asynchronous queues for coroutines. These classes are very similar
+to those provided in the standard library's `asyncio package
+<https://docs.python.org/3/library/asyncio-queue.html>`_.
+
+... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/queues.py |
Fill in missing docstrings in my code | #
# Copyright 2011 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,7 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""Miscellaneous network utility code."""
import asyncio
import concurrent.futures
@@ -60,6 +61,29 @@ flags: Optional[int] = None,
reuse_port: bool = False,
) -> List[socket.socket]... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/netutil.py |
Replace inline comments with docstrings | #
# Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,17 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""Escaping/unescaping methods for HTML, JSON, URLs, and others.
+
+Also includes a few other miscellaneous string manipulation functions that
+have crept in over time.
+
+Many functions in this... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/escape.py |
Write docstrings for this repository | #
# Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,48 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""``tornado.web`` provides a simple web framework with asynchronous
+features that allow it to scale to large numbers of open connections,
+making it ideal for `long polling
+<http://en.wikiped... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/web.py |
Create documentation for each function signature | # Copyright 2015 The Tornado Authors
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in w... | --- +++ @@ -29,6 +29,14 @@
class _TimeoutGarbageCollector:
+ """Base class for objects that periodically clean up timed-out waiters.
+
+ Avoids memory leak in a common pattern like:
+
+ while True:
+ yield condition.wait(short_timeout)
+ print('looping....')
+ """
def __... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/locks.py |
Fully document this Python code with docstrings | #
# Copyright 2012 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -12,6 +12,21 @@ # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
+"""Logging support for Tornado.
+
+Tornado uses three logger streams:
+
+* ``tornado.access``: Per-request logging for T... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/log.py |
Add verbose docstrings with examples | #
# Copyright 2014 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,7 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""A non-blocking TCP connection factory."""
import functools
import socket
@@ -37,6 +38,22 @@
class _Connector:
+ """A stateless implementation of the "Happy Eyeballs" algorithm.
+
+ ... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/tcpclient.py |
Add docstrings that explain inputs and outputs | #
# Copyright 2012 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -12,6 +12,18 @@ # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
+"""Utilities for working with ``Future`` objects.
+
+Tornado previously provided its own ``Future`` class, but now uses
... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/concurrent.py |
Add professional docstrings to my codebase | #
# Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,11 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""HTTP utility code shared by clients and servers.
+
+This module also defines the `HTTPServerRequest` class which is exposed
+via `tornado.web.RequestHandler.request`.
+"""
import calendar
... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/httputil.py |
Add return value explanations in docstrings | #
# Copyright 2011 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, s... | --- +++ @@ -13,6 +13,7 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""A non-blocking, single-threaded TCP server."""
import errno
import os
@@ -40,6 +41,85 @@
class TCPServer:
+ r"""A non-blocking, single-threaded TCP server.
+
+ To use `TCPServer`... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/tcpserver.py |
Write documentation strings for class attributes |
import array
import asyncio
from inspect import getfullargspec
import os
import re
import typing
import zlib
from typing import (
Any,
Optional,
Dict,
Mapping,
List,
Tuple,
Match,
Callable,
Type,
Sequence,
)
if typing.TYPE_CHECKING:
# Additional imports only used in type c... | --- +++ @@ -1,3 +1,14 @@+"""Miscellaneous utility functions and classes.
+
+This module is used internally by Tornado. It is not necessarily expected
+that the functions and classes defined here will be useful to other
+applications, but they are documented here in case they are.
+
+The one public-facing part of this ... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/util.py |
Write documentation strings for class attributes | # Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, sof... | --- +++ @@ -12,6 +12,30 @@ # License for the specific language governing permissions and limitations
# under the License.
+"""Translation methods for generating localized strings.
+
+To load a locale and generate a translated string::
+
+ user_locale = tornado.locale.get("es_LA")
+ print(user_locale.translate(... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/locale.py |
Help me write clear docstrings | # Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under th... | --- +++ @@ -9,6 +9,7 @@ # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
+"""Bridges between the Twisted package and Tornado."""
import sys
@@ -22,6 +23,24 @@
def install() -> None:
+ "... | https://raw.githubusercontent.com/tornadoweb/tornado/HEAD/tornado/platform/twisted.py |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.