repo stringlengths 7 90 | file_url stringlengths 81 315 | file_path stringlengths 4 228 | content stringlengths 0 32.8k | language stringclasses 1
value | license stringclasses 7
values | commit_sha stringlengths 40 40 | retrieved_at stringdate 2026-01-04 14:38:15 2026-01-05 02:33:18 | truncated bool 2
classes |
|---|---|---|---|---|---|---|---|---|
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/inference/hybridengine/polices/llama.py | colossalai/legacy/inference/hybridengine/polices/llama.py | from functools import partial
from typing import List
import torch
from torch.nn import Module
from transformers.models.llama.modeling_llama import (
LlamaAttention,
LlamaDecoderLayer,
LlamaForCausalLM,
LlamaModel,
LlamaRMSNorm,
)
from colossalai.shardformer.policies.base_policy import ModulePolic... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/inference/hybridengine/polices/__init__.py | colossalai/legacy/inference/hybridengine/polices/__init__.py | from .llama import LlamaModelInferPolicy
__all__ = ["LlamaModelInferPolicy"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/inference/pipeline/microbatch_manager.py | colossalai/legacy/inference/pipeline/microbatch_manager.py | from enum import Enum
from typing import Dict
import torch
from ..tensor_parallel.batch_infer_state import BatchInferState
from ..tensor_parallel.kvcache_manager import MemoryManager
__all__ = "MicroBatchManager"
class Status(Enum):
PREFILL = 1
GENERATE = 2
DONE = 3
COOLDOWN = 4
class MicroBatchD... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/inference/pipeline/__init__.py | colossalai/legacy/inference/pipeline/__init__.py | from .microbatch_manager import MicroBatchManager
__all__ = ["MicroBatchManager"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/inference/pipeline/benchmark/benchmark.py | colossalai/legacy/inference/pipeline/benchmark/benchmark.py | import argparse
import time
import torch
import torch.distributed as dist
import transformers
import colossalai
from colossalai.inference import PPInferEngine
from colossalai.inference.pipeline.policies import LlamaModelInferPolicy
GIGABYTE = 1024**3
MEGABYTE = 1024 * 1024
colossalai.launch_from_torch()
def data_... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/registry/registry.py | colossalai/legacy/registry/registry.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
from types import ModuleType
from typing import List
class Registry:
"""This is a registry class used to register classes and modules so that a universal
object builder can be enabled.
Args:
name (str): The name of the registry .
third_part... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/registry/__init__.py | colossalai/legacy/registry/__init__.py | import torch.distributed.optim as dist_optim
import torch.nn as nn
import torch.optim as optim
from .registry import Registry
LAYERS = Registry("layers", third_party_library=[nn])
MODELS = Registry("models")
OPTIMIZERS = Registry("optimizers", third_party_library=[optim, dist_optim])
DATASETS = Registry("datasets")
D... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/parallel_context.py | colossalai/legacy/context/parallel_context.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import random
import socket
from collections import Counter
from typing import Union
import numpy as np
import torch
import torch.distributed as dist
from colossalai.context.config import Config
from colossalai.context.singleton_meta import SingletonMeta
from colossala... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/parallel_mode.py | colossalai/legacy/context/parallel_mode.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
from enum import Enum
# parallel modes
class ParallelMode(Enum):
"""This is an enumeration class containing all possible parallel modes."""
GLOBAL = "global"
# common parallel
DATA = "data"
# model parallel - containing tensor and pipeline parall... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/__init__.py | colossalai/legacy/context/__init__.py | from .parallel_context import ParallelContext
from .parallel_mode import ParallelMode
from .process_group_initializer import *
from .random import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/process_group_initializer.py | colossalai/legacy/context/process_group_initializer/process_group_initializer.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
from abc import ABC, abstractmethod
from colossalai.context import Config
class ProcessGroupInitializer(ABC):
"""An object, knowing the parallelism configuration, that initializes parallel groups.
Args:
rank (int): The rank of current process.
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_data.py | colossalai/legacy/context/process_group_initializer/initializer_data.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
from torch import distributed as dist
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitializer
@DIST_GROUP_INITIALIZER.register_module
class Initializer_Data(... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_2p5d.py | colossalai/legacy/context/process_group_initializer/initializer_2p5d.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import math
import torch.distributed as dist
from colossalai.context import Config
from colossalai.legacy.global_variables import tensor_parallel_env as env
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .pr... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_3d.py | colossalai/legacy/context/process_group_initializer/initializer_3d.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import math
import torch.distributed as dist
from colossalai.legacy.global_variables import tensor_parallel_env as env
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import Process... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_1d.py | colossalai/legacy/context/process_group_initializer/initializer_1d.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import torch.distributed as dist
from colossalai.legacy.global_variables import tensor_parallel_env as env
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitiali... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_sequence.py | colossalai/legacy/context/process_group_initializer/initializer_sequence.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import torch.distributed as dist
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .initializer_tensor import Initializer_Tensor
from .process_group_initializer import ProcessGroupInitializer
@DIST_GROUP_INITIA... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_model.py | colossalai/legacy/context/process_group_initializer/initializer_model.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import torch.distributed as dist
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitializer
@DIST_GROUP_INITIALIZER.register_module
class Initializer_Model(Proc... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_2d.py | colossalai/legacy/context/process_group_initializer/initializer_2d.py | import math
import torch.distributed as dist
from colossalai.legacy.global_variables import tensor_parallel_env as env
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitializer
def _check_summa_env_var(summa... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/__init__.py | colossalai/legacy/context/process_group_initializer/__init__.py | from .initializer_1d import Initializer_1D
from .initializer_2d import Initializer_2D
from .initializer_2p5d import Initializer_2p5D
from .initializer_3d import Initializer_3D
from .initializer_data import Initializer_Data
from .initializer_model import Initializer_Model
from .initializer_pipeline import Initializer_Pi... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_pipeline.py | colossalai/legacy/context/process_group_initializer/initializer_pipeline.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
from torch import distributed as dist
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitializer
@DIST_GROUP_INITIALIZER.register_module
class Initializer_Pipel... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/process_group_initializer/initializer_tensor.py | colossalai/legacy/context/process_group_initializer/initializer_tensor.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import torch.distributed as dist
from colossalai.legacy.registry import DIST_GROUP_INITIALIZER
from ..parallel_mode import ParallelMode
from .process_group_initializer import ProcessGroupInitializer
@DIST_GROUP_INITIALIZER.register_module
class Initializer_Tensor(Pro... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/random/seed_manager.py | colossalai/legacy/context/random/seed_manager.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import torch
from torch import Tensor
from colossalai.legacy.context.parallel_mode import ParallelMode
class SeedManager:
"""This class is a manager of all random seeds involved in the system.
Note:
The parallel_mode should be concluded in ``ParallelM... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/random/_helper.py | colossalai/legacy/context/random/_helper.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import functools
from contextlib import contextmanager
import torch.cuda
from torch import Tensor
from ..parallel_mode import ParallelMode
from .seed_manager import SeedManager
_SEED_MANAGER = SeedManager()
def get_seeds():
"""Returns the seeds of the seed manag... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/context/random/__init__.py | colossalai/legacy/context/random/__init__.py | from ._helper import (
add_seed,
get_current_mode,
get_seeds,
get_states,
moe_set_seed,
reset_seeds,
seed,
set_mode,
set_seed_states,
sync_states,
with_seed,
)
__all__ = [
"seed",
"set_mode",
"with_seed",
"add_seed",
"get_seeds",
"get_states",
"ge... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/load_balance.py | colossalai/legacy/moe/load_balance.py | from copy import deepcopy
from typing import List, Optional, Tuple
import torch
import torch.distributed as dist
from torch import Tensor, nn
from torch.distributed import ProcessGroup
from colossalai.cluster import ProcessGroupMesh
from colossalai.legacy.moe.manager import MOE_MANAGER
from colossalai.shardformer.lay... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/utils.py | colossalai/legacy/moe/utils.py | import contextlib
import os
from typing import Any, Callable, Dict, List, Optional, Tuple
import torch
import torch.distributed as dist
import torch.nn as nn
import torch.nn.functional as F
from torch.distributed.distributed_c10d import get_process_group_ranks
from colossalai.accelerator import get_accelerator
from c... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/manager.py | colossalai/legacy/moe/manager.py | from typing import Tuple
import torch
import torch.distributed as dist
from colossalai.context.singleton_meta import SingletonMeta
from colossalai.tensor.moe_tensor.api import get_moe_info
from colossalai.tensor.moe_tensor.moe_info import MoeParallelInfo
class MoEManager(metaclass=SingletonMeta):
"""MoE manager... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/train.py | colossalai/legacy/moe/openmoe/train.py | import argparse
import os
from functools import partial
from typing import Dict
import torch
import torch.distributed as dist
from datasets import load_dataset
from huggingface_hub import snapshot_download
from model.modeling_openmoe import OpenMoeForCausalLM, set_openmoe_args
from model.openmoe_policy import OpenMoeF... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/infer.py | colossalai/legacy/moe/openmoe/infer.py | from argparse import ArgumentParser
import torch
from model.modeling_openmoe import OpenMoeForCausalLM, set_openmoe_args
from transformers import T5Tokenizer
from transformers.models.llama import LlamaConfig
def parse_args():
parser = ArgumentParser()
parser.add_argument("--model", default="base", type=str, ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/model/openmoe_policy.py | colossalai/legacy/moe/openmoe/model/openmoe_policy.py | from functools import partial
from typing import Callable, Dict, List, Optional, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch import Tensor
from torch.nn import Module
from transformers.modeling_outputs import CausalLMOutputWithPast
from transformers.utils import logging
from co... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/model/convert_openmoe_ckpt.py | colossalai/legacy/moe/openmoe/model/convert_openmoe_ckpt.py | # coding=utf-8
# Copyright 2022 Google LLC and HuggingFace Inc. team.
#
# 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... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/model/__init__.py | colossalai/legacy/moe/openmoe/model/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/model/modeling_openmoe.py | colossalai/legacy/moe/openmoe/model/modeling_openmoe.py | # coding=utf-8
# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to G... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | true |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/benchmark/benchmark_fsdp.py | colossalai/legacy/moe/openmoe/benchmark/benchmark_fsdp.py | import argparse
import functools
import os
import torch
import torch.distributed as dist
import tqdm
from model.modeling_openmoe import LlamaConfig, OpenMoeDecoderLayer, OpenMoeForCausalLM, set_openmoe_args
from torch.distributed.fsdp import FullyShardedDataParallel as FSDP
from torch.distributed.fsdp.fully_sharded_da... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/benchmark/utils.py | colossalai/legacy/moe/openmoe/benchmark/utils.py | from time import time
from typing import Optional
import torch
import torch.distributed as dist
import torch.nn as nn
from torch import Tensor
from colossalai.logging import DistributedLogger
def print_model_numel(logger: DistributedLogger, model: nn.Module) -> None:
B = 1024**3
M = 1024**2
K = 1024
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/openmoe/benchmark/benchmark_cai.py | colossalai/legacy/moe/openmoe/benchmark/benchmark_cai.py | import argparse
import json
import os
import torch
import torch.distributed as dist
from huggingface_hub import snapshot_download
from model.modeling_openmoe import OpenMoeForCausalLM, set_openmoe_args
from model.openmoe_policy import OpenMoeForCausalLMPolicy
from torch.utils.data import Dataset
from tqdm import tqdm
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/layer/experts.py | colossalai/legacy/moe/layer/experts.py | import math
from typing import Callable, Optional, Tuple
import torch
import torch.nn as nn
from colossalai.kernel.triton.llama_act_combine_kernel import HAS_TRITON
from colossalai.legacy.moe.manager import MOE_MANAGER
from colossalai.legacy.moe.utils import get_activation
from colossalai.moe._operation import EPGrad... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/layer/__init__.py | colossalai/legacy/moe/layer/__init__.py | from .experts import *
from .layers import *
from .routers import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/layer/routers.py | colossalai/legacy/moe/layer/routers.py | import math
from typing import Callable, Optional, Tuple
import torch
import torch.nn as nn
from colossalai.kernel.triton.llama_act_combine_kernel import HAS_TRITON
from colossalai.legacy.moe.manager import MOE_MANAGER
from colossalai.legacy.moe.utils import get_activation
from colossalai.moe._operation import EPGrad... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/moe/layer/layers.py | colossalai/legacy/moe/layer/layers.py | import dataclasses
import math
from typing import Any, Optional, Tuple
import torch
import torch.distributed as dist
import torch.nn as nn
import torch.nn.functional as F
from colossalai.legacy.moe.load_balance import LoadBalancer
from colossalai.legacy.moe.utils import create_ep_hierarchical_group, get_noise_generat... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/pipeline_process_group.py | colossalai/legacy/pipeline/pipeline_process_group.py | import threading
from typing import List
import torch.distributed as dist
from torch.distributed import rpc
from colossalai.legacy.tensor import ProcessGroup
class PipelineProcessGroup:
# TODO : flexible API for DP size and TP size
# In the future design mode, dp_degree and tp_degree should be removed
d... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/pipelinable.py | colossalai/legacy/pipeline/pipelinable.py | import torch
from colossalai.legacy.context import ParallelMode
from colossalai.legacy.core import global_context as gpc
from colossalai.legacy.nn.layer.utils import CheckpointModule
from colossalai.tensor import ColoParameter
from colossalai.utils.model.utils import InsertPostInitMethodToModuleSubClasses
from .layer... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/utils.py | colossalai/legacy/pipeline/utils.py | import heapq
import inspect
from collections import OrderedDict
from typing import List
import torch
from colossalai.legacy.nn.layer.utils import CheckpointModule
from colossalai.logging import get_dist_logger
def _binary_partition(weights: List, start: int, end: int):
"""Returns the binary partition position o... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/__init__.py | colossalai/legacy/pipeline/__init__.py | from .layer_spec import LayerSpec
from .pipelinable import PipelinableContext, PipelinableModel
__all__ = ["PipelinableModel", "PipelinableContext", "LayerSpec"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/layer_spec.py | colossalai/legacy/pipeline/layer_spec.py | import torch
from colossalai.utils.model.utils import call_to_str
class LayerSpec:
""" """
def __init__(self, typename, *module_args, **module_kwargs):
self.typename = typename
self.module_args = module_args
self.module_kwargs = module_kwargs
self.children = None
self... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/middleware/topo.py | colossalai/legacy/pipeline/middleware/topo.py | from dataclasses import dataclass
from typing import Dict, List
# This file includes data structure used by Pipeline Middleware.
@dataclass
class ValPosition:
partition_id: int
offset: int
def __str__(self) -> str:
res = f"[partition_id:{self.partition_id},offset:{self.offset}]"
return r... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/middleware/__init__.py | colossalai/legacy/pipeline/middleware/__init__.py | from .topo import Partition, PartitionInputVal, PartitionOutputVal, Topo
__all__ = ["Topo", "Partition", "PartitionOutputVal", "PartitionInputVal"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/middleware/adaptor/fx.py | colossalai/legacy/pipeline/middleware/adaptor/fx.py | import torch
from torch.fx.graph_module import GraphModule
from colossalai.legacy.pipeline.middleware.topo import Partition, PartitionInputVal, PartitionOutputVal, Topo
def partition_name_to_id(partition_name, is_input=False, is_output=False):
if is_input:
partition_id = 0
elif is_output:
par... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/middleware/adaptor/__init__.py | colossalai/legacy/pipeline/middleware/adaptor/__init__.py | from .fx import get_topology as get_fx_topology
__all__ = ["get_fx_topology"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/rpc/_pipeline_schedule.py | colossalai/legacy/pipeline/rpc/_pipeline_schedule.py | import threading
from typing import Callable, Dict, List
import torch
from torch._C._distributed_rpc import PyRRef
from torch.futures import Future
from colossalai.legacy.pipeline.pipeline_process_group import ppg
from colossalai.legacy.pipeline.rpc._pipeline_base import Phase, PipelineEngineBase, UniqueKey, WorkerBa... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/rpc/utils.py | colossalai/legacy/pipeline/rpc/utils.py | import argparse
import os
import warnings
from typing import Any, Callable, Tuple, Type, Union
import torch
import torch.distributed.rpc as rpc
import torch.multiprocessing as mp
from torch._C._distributed_rpc import _is_current_rpc_agent_set
from torch.futures import Future
from colossalai.initialize import launch
f... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/rpc/_pipeline_base.py | colossalai/legacy/pipeline/rpc/_pipeline_base.py | import inspect
import math
import threading
from abc import ABC, abstractmethod
from enum import Enum
from functools import partial
from typing import Any, Callable, Dict, List, Tuple
import torch
import torch.distributed.rpc as rpc
from torch import autograd, nn, optim
from torch._C._distributed_rpc import PyRRef
fro... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | true |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/legacy/pipeline/rpc/__init__.py | colossalai/legacy/pipeline/rpc/__init__.py | from ._pipeline_schedule import ChimeraPipelineEngine, FillDrainPipelineEngine, OneFOneBPipelineEngine
from .utils import pytree_map
__all__ = ["FillDrainPipelineEngine", "OneFOneBPipelineEngine", "ChimeraPipelineEngine", "pytree_map"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/__init__.py | colossalai/nn/__init__.py | from .init import *
from .layer import *
from .loss import *
from .lr_scheduler import *
from .optimizer import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/init.py | colossalai/nn/init.py | import math
import warnings
import torch.nn as nn
from torch import Tensor
def zeros_():
"""Return the initializer filling the input Tensor with the scalar zeros"""
def initializer(tensor: Tensor, fan_in: int = None, fan_out: int = None):
return nn.init.zeros_(tensor)
return initializer
def o... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/loss/__init__.py | colossalai/nn/loss/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/layer/layernorm.py | colossalai/nn/layer/layernorm.py | """This code is from NVIDIA apex:
https://github.com/NVIDIA/apex
with some changes. """
import numbers
import torch
from torch.cuda.amp import custom_bwd, custom_fwd
from torch.nn import init
from torch.nn.parameter import Parameter
from colossalai.kernel.kernel_loader import LayerNormLoader
try:
from ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/layer/utils.py | colossalai/nn/layer/utils.py | def divide(numerator, denominator):
"""Only allow exact division.
Args:
numerator (int): Numerator of the division.
denominator (int): Denominator of the division.
Returns:
int: the result of exact division.
"""
assert denominator != 0, "denominator can not be zero"
ass... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/layer/__init__.py | colossalai/nn/layer/__init__.py | from .utils import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/layer/scaled_softmax.py | colossalai/nn/layer/scaled_softmax.py | # This code from NVIDIA Megatron:
# with minor changes.
import enum
import torch
import torch.nn as nn
from colossalai.kernel.kernel_loader import ScaledMaskedSoftmaxLoader, ScaledUpperTriangleMaskedSoftmaxLoader
# NOTE: These kernels are compiled on specific GPU arch and not widely applicable.
# try:
# fro... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/onecycle.py | colossalai/nn/lr_scheduler/onecycle.py | from torch.optim.lr_scheduler import OneCycleLR as _OneCycleLR
class OneCycleLR(_OneCycleLR):
r"""Sets the learning rate of each parameter group according to the
1cycle learning rate policy. The 1cycle policy anneals the learning
rate from an initial learning rate to some maximum learning rate and then
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/multistep.py | colossalai/nn/lr_scheduler/multistep.py | from typing import List
from torch.optim.lr_scheduler import MultiStepLR as _MultiStepLR
from .delayed import WarmupScheduler
class MultiStepLR(_MultiStepLR):
"""Decays the learning rate of each parameter group by gamma once the
number of epoch reaches one of the milestones. Notice that such decay can
h... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/delayed.py | colossalai/nn/lr_scheduler/delayed.py | import torch
from packaging.version import Version
if Version(torch.__version__) >= Version("2.0.0"):
from torch.optim.lr_scheduler import LRScheduler as _LRScheduler
else:
from torch.optim.lr_scheduler import _LRScheduler
from colossalai.logging import get_dist_logger
class _enable_get_lr_call:
def __i... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/__init__.py | colossalai/nn/lr_scheduler/__init__.py | from .cosine import CosineAnnealingLR, CosineAnnealingWarmupLR, FlatAnnealingLR, FlatAnnealingWarmupLR
from .linear import LinearWarmupLR
from .multistep import MultiStepLR, MultiStepWarmupLR
from .onecycle import OneCycleLR
from .poly import PolynomialLR, PolynomialWarmupLR
from .torch import ExponentialLR, LambdaLR, ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/cosine.py | colossalai/nn/lr_scheduler/cosine.py | from torch.optim.lr_scheduler import CosineAnnealingLR as _CosineAnnealingLR
from .delayed import DelayerScheduler, WarmupDelayerScheduler, WarmupScheduler
class CosineAnnealingLR(_CosineAnnealingLR):
r"""Set the learning rate of each parameter group using a cosine annealing
schedule, where :math:`\eta_{max}... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/poly.py | colossalai/nn/lr_scheduler/poly.py | from torch.optim.lr_scheduler import _LRScheduler
from .delayed import WarmupScheduler
class PolynomialLR(_LRScheduler):
"""Polynomial learning rate scheduler.
Args:
optimizer (:class:`torch.optim.Optimizer`): Wrapped optimizer.
total_steps (int): Number of total training steps.
end_... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/torch.py | colossalai/nn/lr_scheduler/torch.py | from torch.optim.lr_scheduler import ExponentialLR as _ExponentialLR
from torch.optim.lr_scheduler import LambdaLR as _LambdaLR
from torch.optim.lr_scheduler import MultiplicativeLR as _MultiplicativeLR
from torch.optim.lr_scheduler import StepLR as _StepLR
class LambdaLR(_LambdaLR):
"""Sets the learning rate of ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/lr_scheduler/linear.py | colossalai/nn/lr_scheduler/linear.py | from torch.optim.lr_scheduler import _LRScheduler
class LinearWarmupLR(_LRScheduler):
"""Linearly warmup learning rate and then linearly decay.
Args:
optimizer (:class:`torch.optim.Optimizer`): Wrapped optimizer.
total_steps (int): Number of total training steps.
warmup_steps (int, op... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/distributed_lamb.py | colossalai/nn/optimizer/distributed_lamb.py | # Disclaimer: Modified from https://github.com/NUS-HPC-AI-Lab/pytorch-lamb/blob/master/optim/lamb.py
from typing import Dict, Optional
import torch
import torch.distributed as dist
from colossalai.interface.optimizer import DistributedOptim
from colossalai.tensor.d_tensor import is_distributed_tensor
__all__ = ["D... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/fused_lamb.py | colossalai/nn/optimizer/fused_lamb.py | # modified from https://github.com/NVIDIA/apex/blob/master/apex/optimizers/fused_lamb.py
import torch
from colossalai.utils import multi_tensor_applier
class FusedLAMB(torch.optim.Optimizer):
"""Implements LAMB algorithm.
`FusedLAMB` requires CUDA extensions which can be built during installation or runtime... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/hybrid_adam.py | colossalai/nn/optimizer/hybrid_adam.py | from typing import Any, Optional
import torch
from colossalai.kernel.kernel_loader import FusedOptimizerLoader
from colossalai.utils import get_current_device, multi_tensor_applier
from .cpu_adam import CPUAdam
class HybridAdam(CPUAdam):
"""Implements Adam algorithm.
Supports parameters updating on both G... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/distributed_adafactor.py | colossalai/nn/optimizer/distributed_adafactor.py | import math
from typing import Dict
import torch
import torch.distributed as dist
from colossalai.interface.optimizer import DistributedOptim
from colossalai.shardformer.layer._operation import _gather, _split
from colossalai.tensor.d_tensor import get_sharding_spec, is_distributed_tensor
# DistributedAdaFactor (wit... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/lars.py | colossalai/nn/optimizer/lars.py | """Adapted from https://github.com/NUS-HPC-AI-Lab/LARS-ImageNet-PyTorch/blob/main/lars.py"""
from typing import Iterable
import torch
from torch.optim import Optimizer
class Lars(Optimizer):
r"""Implements the LARS optimizer from `"Large batch training of convolutional networks"
<https://arxiv.org/pdf/1708.... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/fused_adam.py | colossalai/nn/optimizer/fused_adam.py | # modified from https://github.com/NVIDIA/apex/blob/master/apex/optimizers/fused_adam.py
"""
Copyright 2020 The Microsoft DeepSpeed Team
Copyright NVIDIA/apex
This file is adapted from fused adam in NVIDIA/apex, commit a109f85
Licensed under the MIT License.
"""
import torch
from colossalai.utils import get_current_d... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/came.py | colossalai/nn/optimizer/came.py | # Copied from https://github.com/yangluo7/CAME/blob/master/came_pytorch/CAME.py
import torch
import torch.optim
class CAME(torch.optim.Optimizer):
"""Implements CAME algorithm.
This implementation is based on:
`CAME: Confidence-guided Adaptive Memory Efficient Optimization`
Args:
params (itera... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/galore.py | colossalai/nn/optimizer/galore.py | """ adapted from https://github.com/jiaweizzhao/GaLore/blob/master/galore_torch/adamw8bit.py"""
import warnings
from typing import List
import torch
from bitsandbytes.optim.optimizer import Optimizer2State
from torch._C import _LinAlgError
def get_galore_param_groups(
model, weight_decay, rank=256, update_proj_... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/__init__.py | colossalai/nn/optimizer/__init__.py | from galore_torch import GaLoreAdafactor, GaLoreAdamW
from colossalai.logging import get_dist_logger
from .came import CAME
from .cpu_adam import CPUAdam
from .distributed_adafactor import DistributedAdaFactor
from .distributed_came import DistributedCAME
from .distributed_galore import DistGaloreAwamW
from .distribu... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/adafactor.py | colossalai/nn/optimizer/adafactor.py | # coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
#
# 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/LICEN... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/distributed_came.py | colossalai/nn/optimizer/distributed_came.py | from typing import Dict
import torch
import torch.distributed as dist
from colossalai.interface.optimizer import DistributedOptim
from colossalai.shardformer.layer._operation import _gather, _split
from colossalai.tensor.d_tensor import get_sharding_spec, is_distributed_tensor
class DistributedCAME(DistributedOptim... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/lamb.py | colossalai/nn/optimizer/lamb.py | """
Adapted from the pytorch-lamb library at https://github.com/cybertronai/pytorch-lamb
"""
import torch
from torch.optim import Optimizer
class Lamb(Optimizer):
r"""Implements Lamb algorithm.
It has been proposed in `Large Batch Optimization for Deep Learning: Training BERT in 76 minutes`_.
Arguments:... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/nvme_optimizer.py | colossalai/nn/optimizer/nvme_optimizer.py | import math
import os
import tempfile
from typing import Callable, Dict, List, Optional
import torch
from torch.nn.parameter import Parameter
class NVMeOptimizer(torch.optim.Optimizer):
"""A base class for offloading optimizer states.
Args:
params: parameters
defaults (dict): default dict
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/distributed_galore.py | colossalai/nn/optimizer/distributed_galore.py | """ adapted from https://github.com/jiaweizzhao/GaLore/blob/master/galore_torch/adamw8bit.py"""
import warnings
from collections import defaultdict
from typing import Dict, Optional
import torch
import torch.distributed as dist
import torch.nn.functional as F
from bitsandbytes.optim.optimizer import Optimizer2State
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/cpu_adam.py | colossalai/nn/optimizer/cpu_adam.py | import math
from typing import Optional
import torch
from colossalai.kernel.kernel_loader import CPUAdamLoader
from .nvme_optimizer import NVMeOptimizer
class CPUAdam(NVMeOptimizer):
"""
Implements Adam algorithm.
Supports parameters updating on both GPU and CPU, depending on the device of parameters.... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/nn/optimizer/fused_sgd.py | colossalai/nn/optimizer/fused_sgd.py | # modified from https://github.com/NVIDIA/apex/blob/master/apex/optimizers/fused_sgd.py
import torch
from torch.optim.optimizer import Optimizer, required
from colossalai.utils import multi_tensor_applier
class FusedSGD(Optimizer):
r"""Implements stochastic gradient descent (optionally with momentum).
`Fuse... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/__init__.py | colossalai/auto_parallel/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/shard_metainfo.py | colossalai/auto_parallel/meta_profiler/shard_metainfo.py | from typing import Callable, List
import torch
from colossalai.auto_parallel.tensor_shard.sharding_strategy import OperationData, ShardingStrategy, TrainCycleItem
from colossalai.tensor.sharding_spec import ShardingSpec
from .constants import INPLACE_MODULE, INPLACE_OPS, NO_SAVE_ACTIVATION
from .registry import meta... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/registry.py | colossalai/auto_parallel/meta_profiler/registry.py | __all__ = ["Registry"]
class Registry:
def __init__(self, name):
self.name = name
self.store = {}
def register(self, source):
def wrapper(func):
if isinstance(source, (list, tuple)):
# support register a list of items for this func
for eleme... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/constants.py | colossalai/auto_parallel/meta_profiler/constants.py | import operator
import torch
import torch.nn as nn
# list of inplace module
INPLACE_MODULE = [nn.ReLU]
# list of inplace operations
INPLACE_OPS = [torch.flatten]
# list of operations that do not save forward activations
NO_SAVE_ACTIVATION = [torch.add, torch.sub, operator.add, operator.sub]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/__init__.py | colossalai/auto_parallel/meta_profiler/__init__.py | from .meta_registry import *
from .registry import meta_register
from .shard_metainfo import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/activation.py | colossalai/auto_parallel/meta_profiler/meta_registry/activation.py | from typing import Callable, List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import ewise_flop_counter as elementwise_flop_counter
from colossalai._analyzer.fx.node_util import compute_size_in_bytes as activation_size
from colossalai.auto_parallel.tensor_shard.sharding_strategy import Memor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/where.py | colossalai/auto_parallel/meta_profiler/meta_registry/where.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes as activation_size
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, TrainCycleItem
from ..registry impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/binary_elementwise_ops.py | colossalai/auto_parallel/meta_profiler/meta_registry/binary_elementwise_ops.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes as activation_size
from colossalai.auto_parallel.tensor_shard.constants import BCAST_FUNC_OP
from colossalai.auto_parallel.tensor_shard.... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/conv.py | colossalai/auto_parallel/meta_profiler/meta_registry/conv.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, OperationDataType, TrainCycleItem
from ..registry impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/pooling.py | colossalai/auto_parallel/meta_profiler/meta_registry/pooling.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, OperationDataType, TrainCycleItem
from ..registry impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/non_spmd.py | colossalai/auto_parallel/meta_profiler/meta_registry/non_spmd.py | import operator
from typing import List, Tuple
import torch
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, TrainCycleItem
from ..registry import meta_register
__all__ = ["non_spmd_meta_info"]
@meta_register.register(torch.Size)
@meta_register.register(torch.Tensor.size)
@meta_regi... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/norm.py | colossalai/auto_parallel/meta_profiler/meta_registry/norm.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, OperationDataType, TrainCycleItem
from ..registry impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/__init__.py | colossalai/auto_parallel/meta_profiler/meta_registry/__init__.py | from .activation import *
from .binary_elementwise_ops import *
from .conv import *
from .embedding import *
from .linear import *
from .non_spmd import *
from .norm import *
from .pooling import *
from .tensor import *
from .where import *
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/embedding.py | colossalai/auto_parallel/meta_profiler/meta_registry/embedding.py | from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, OperationDataType, TrainCycleItem
from ..registry impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/tensor.py | colossalai/auto_parallel/meta_profiler/meta_registry/tensor.py | from typing import Callable, List, Tuple
import torch
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, OperationDataType, TrainCycleItem
from ..registry import meta_register
__all__ = ["tensor_related_metainfo"]
def... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/auto_parallel/meta_profiler/meta_registry/linear.py | colossalai/auto_parallel/meta_profiler/meta_registry/linear.py | from functools import reduce
from typing import List, Tuple
import torch
from colossalai._analyzer._subclasses.flop_tensor import flop_mapping
from colossalai._analyzer.fx.node_util import compute_size_in_bytes
from colossalai.auto_parallel.tensor_shard.sharding_strategy import MemoryCost, TrainCycleItem
from ..regi... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.