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/applications/ColossalChat/coati/models/critic.py | applications/ColossalChat/coati/models/critic.py | """
Critic model
"""
from typing import Optional
import torch
import torch.nn as nn
from coati.models import BaseModel
from transformers import PretrainedConfig
class Critic(BaseModel):
"""
Critic model class.
Args:
pretrained (str): path to pretrained model.
config (PretrainedConfig): ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/utils.py | applications/ColossalChat/coati/models/utils.py | import json
import os
from typing import Any, Dict, Optional, Union
import torch
import torch.nn.functional as F
def get_model_numel(model: torch.nn.Module) -> int:
return sum(p.numel() for p in model.parameters())
def compute_reward(
r: Union[torch.Tensor, float],
kl_coef: float,
log_probs: torch.... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/loss.py | applications/ColossalChat/coati/models/loss.py | """
loss functions
"""
from typing import Optional, Tuple
import torch
import torch.distributed as dist
import torch.nn as nn
from .utils import masked_mean
class GPTLMLoss(nn.Module):
"""
GPT Language Model Loss
"""
def __init__(self):
super().__init__()
# NOTE: default ignore_ind... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/__init__.py | applications/ColossalChat/coati/models/__init__.py | from .base import BaseModel
from .critic import Critic
from .generation import generate, generate_streaming, prepare_inputs_fn, update_model_kwargs_fn
from .lora import LoraConfig, convert_to_lora_module, lora_manager
from .loss import DpoLoss, KTOLoss, LogExpLoss, LogSigLoss, PolicyLoss, ValueLoss
from .reward_model i... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/base.py | applications/ColossalChat/coati/models/base.py | """
Base class for critic and reward model
"""
from typing import Optional
import torch
import torch.nn as nn
from transformers import AutoModel, PretrainedConfig
class BaseModel(nn.Module):
"""
Actor model base class.
Args:
pretrained (str): path to pretrained model.
config (Pretrained... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/generation.py | applications/ColossalChat/coati/models/generation.py | import copy
from typing import Any, Callable, List, Optional
import torch
import torch.distributed as dist
from transformers import PreTrainedTokenizer
try:
from transformers.generation_logits_process import (
LogitsProcessorList,
TemperatureLogitsWarper,
TopKLogitsWarper,
TopPLogi... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/models/rlvr_reward_model.py | applications/ColossalChat/coati/models/rlvr_reward_model.py | """
reward model
"""
from typing import Callable, List, Optional
import torch
class RLVRRewardModel:
"""
RLVRReward model class. Support varifiable reward.
Args:
reward_fn_list List: list of reward functions
**kwargs: all other kwargs as in reward functions
"""
def __init__(sel... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/consumer.py | applications/ColossalChat/coati/distributed/consumer.py | from contextlib import nullcontext
from typing import Any, Dict, Optional
import ray
import ray.util.collective as cc
import torch
import torch.distributed as dist
from coati.distributed.profiling_utils import CustomProfiler
from coati.utils import save_checkpoint
from tqdm import tqdm
from transformers import AutoMod... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/launch_zero_bubble.py | applications/ColossalChat/coati/distributed/launch_zero_bubble.py | import copy
import os
import uuid
from typing import Any, Dict, Optional
import ray
from .comm import SharedVariableActor
from .zero_bubble.distributor import Distributor
from .zero_bubble.grpo_consumer import GRPOConsumer
from .zero_bubble.producer import SimpleProducer
ALGO_MAP = {"GRPO": GRPOConsumer, "DAPO": GRP... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/profiling_utils.py | applications/ColossalChat/coati/distributed/profiling_utils.py | import os
import time
class CustomProfiler:
def __init__(self, name, disabled=True):
self.disabled = disabled
if not disabled:
self.name = name
self.pid = os.getpid()
self.file = open(f"{name}.prof", "w")
def _log(self, message):
if self.disabled:
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/producer.py | applications/ColossalChat/coati/distributed/producer.py | import copy
import json
import os
from typing import Any, Dict, Optional
import ray
import ray.util.collective as cc
import torch
import tqdm
import wandb
from coati.dataset import StatefulDistributedSampler
from coati.dataset.loader import RawConversationDataset, collate_fn_grpo
from coati.distributed.profiling_utils... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/utils.py | applications/ColossalChat/coati/distributed/utils.py | import json
import os
from typing import Any, Dict, List
import torch
from filelock import FileLock
from colossalai.shardformer.layer.loss import dist_log_prob
def unbind_batch(batch: Dict[str, torch.Tensor]) -> List[Dict[str, torch.Tensor]]:
batches = []
for k, v in batch.items():
if len(batches) =... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/comm.py | applications/ColossalChat/coati/distributed/comm.py | import copy
from typing import Any, Dict
import ray
import ray.util.collective as cc
import torch
import torch.distributed.distributed_c10d as c10d
from packaging.version import Version
def ray_broadcast_object(obj: Any, src: int = 0, device=None, group_name: str = "default") -> Any:
rank = cc.get_rank(group_nam... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/loss.py | applications/ColossalChat/coati/distributed/loss.py | from typing import Optional
import torch
import torch.nn as nn
from coati.distributed.utils import masked_mean, masked_sum
class PolicyLoss(nn.Module):
"""
Policy Loss for PPO
"""
def __init__(
self,
clip_eps_low: float = 0.2,
clip_eps_high: float = 0.2,
beta: float =... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/grpo_consumer.py | applications/ColossalChat/coati/distributed/grpo_consumer.py | from contextlib import nullcontext
from typing import Any, Optional
import ray
import torch
import wandb
from coati.distributed.consumer import BaseConsumer
from coati.distributed.loss import PolicyLoss
from coati.distributed.utils import entropy_from_logits, memory_efficient_logprob
from coati.trainer.utils import al... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/__init__.py | applications/ColossalChat/coati/distributed/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/launch.py | applications/ColossalChat/coati/distributed/launch.py | import copy
import os
import uuid
from typing import Any, Dict, Optional
import ray
from .consumer import SimpleConsumer
from .grpo_consumer import GRPOConsumer
from .producer import SimpleProducer
ALGO_MAP = {
"Simple": SimpleConsumer,
"GRPO": GRPOConsumer,
"DAPO": GRPOConsumer,
"REINFORCE_PPB": GRP... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/inference_backend.py | applications/ColossalChat/coati/distributed/inference_backend.py | from typing import Any, Dict
import torch
import torch.nn.functional as F
from transformers import AutoConfig, AutoModelForCausalLM, PreTrainedModel, PreTrainedTokenizer
from colossalai.utils import get_current_device
from .utils import log_probs_from_logits, update_by_default
try:
import sglang as sgl
except I... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/reward/reward_utils.py | applications/ColossalChat/coati/distributed/reward/reward_utils.py | # Copyright Unakar
# Modified from https://github.com/Unakar/Logic-RL/blob/086373176ac198c97277ff50f4b6e7e1bfe669d3/verl/utils/reward_score/kk.py#L99
#
# 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 Lice... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/reward/reward_fn.py | applications/ColossalChat/coati/distributed/reward/reward_fn.py | # Copyright 2024 ByteDance Group
# 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,... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/reward/verifiable_reward.py | applications/ColossalChat/coati/distributed/reward/verifiable_reward.py | """
Function-based reward verification module.
"""
import inspect
from typing import Any, Dict, List
import torch
class VerifiableReward:
def __init__(self, reward_fns: List[callable], **kwargs: List[Dict[str, Any]]):
self.reward_fns = reward_fns
self.kwargs = kwargs
def __call__(
s... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/reward/code_reward/testing_util.py | applications/ColossalChat/coati/distributed/reward/code_reward/testing_util.py | # Code from the verl Project (https://github.com/agentica-project/rllm),
# which itself is adapted from Prime (https://github.com/PRIME-RL/PRIME)
#
# Copyright 2024 ByteDance Group
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/reward/code_reward/utils.py | applications/ColossalChat/coati/distributed/reward/code_reward/utils.py | # Code from the verl Project (https://github.com/agentica-project/rllm),
# which itself is adapted from Prime (https://github.com/PRIME-RL/PRIME)
#
# Copyright 2024 ByteDance Group
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/zero_bubble/consumer.py | applications/ColossalChat/coati/distributed/zero_bubble/consumer.py | import os
import threading
import time
from typing import Any, Dict, Optional
import ray
import ray.util.collective as cc
import torch
import torch.distributed as dist
from coati.distributed.comm import SharedVariableActor, ray_broadcast_tensor_dict
from coati.distributed.profiling_utils import CustomProfiler
from coa... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/zero_bubble/producer.py | applications/ColossalChat/coati/distributed/zero_bubble/producer.py | import copy
import json
import os
import threading
import time
from typing import Any, Dict, Optional
import ray
import ray.util.collective as cc
import torch
import tqdm
import wandb
from coati.dataset.loader import RawConversationDataset, collate_fn_grpo
from coati.distributed.comm import SharedVariableActor, ray_br... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/zero_bubble/distributor.py | applications/ColossalChat/coati/distributed/zero_bubble/distributor.py | import time
import ray
import ray.util.collective as cc
import torch
from coati.distributed.comm import SharedVariableActor, ray_broadcast_tensor_dict
from coati.distributed.profiling_utils import CustomProfiler
from colossalai.utils import get_current_device
@ray.remote
class Distributor:
def __init__(
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/zero_bubble/grpo_consumer.py | applications/ColossalChat/coati/distributed/zero_bubble/grpo_consumer.py | from contextlib import nullcontext
from typing import Any, Optional
import ray
import torch
import wandb
from coati.distributed.comm import SharedVariableActor
from coati.distributed.loss import PolicyLoss
from coati.distributed.utils import entropy_from_logits, memory_efficient_logprob
from coati.distributed.zero_bub... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/distributed/zero_bubble/__init__.py | applications/ColossalChat/coati/distributed/zero_bubble/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/ppo.py | applications/ColossalChat/coati/trainer/ppo.py | """
PPO trainer
"""
import os
from typing import Dict, List, Optional, Union
import torch
import wandb
from coati.experience_buffer import NaiveExperienceBuffer
from coati.experience_maker import Experience, NaiveExperienceMaker
from coati.models import Critic, RewardModel, RLVRRewardModel
from coati.models.loss impo... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/orpo.py | applications/ColossalChat/coati/trainer/orpo.py | """
Orpo trainer
"""
import os
from typing import Any, Optional
import torch
from coati.models.loss import OddsRatioLoss
from coati.models.utils import calc_masked_log_probs
from coati.trainer.utils import all_reduce_mean
from coati.utils import AccumulativeMeanMeter, save_checkpoint
from torch.optim import Optimizer... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/rm.py | applications/ColossalChat/coati/trainer/rm.py | """
Reward model trianer
"""
import os
from typing import Any, Callable, Optional
import torch
import tqdm
from coati.models import LogSigLoss
from coati.trainer.utils import all_reduce_mean
from coati.utils import AccumulativeMeanMeter, save_checkpoint
from torch.optim import Optimizer
from torch.optim.lr_scheduler ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/sft.py | applications/ColossalChat/coati/trainer/sft.py | """
SFT trainer
"""
import os
from typing import Optional
import torch
import torch.distributed as dist
from coati.trainer.utils import all_reduce_mean
from coati.utils import AccumulativeMeanMeter, save_checkpoint
from torch.optim import Optimizer
from torch.optim.lr_scheduler import _LRScheduler
from torch.utils.da... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/utils.py | applications/ColossalChat/coati/trainer/utils.py | """
Training utilities for Coati.
"""
from typing import Any
import torch
import torch.distributed as dist
from torch.utils._pytree import tree_map
from torch.utils.data import DataLoader
from colossalai.booster import Plugin
class AnnealingScheduler:
def __init__(self, start, end, warmup_steps=100, annealing_... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/kto.py | applications/ColossalChat/coati/trainer/kto.py | """
KTO trainer
"""
import os
from typing import Any, Optional
import torch
import torch.distributed as dist
from coati.models.loss import KTOLoss
from coati.models.utils import calc_masked_log_probs
from coati.trainer.utils import all_reduce_mean
from coati.utils import AccumulativeMeanMeter, save_checkpoint
from to... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/__init__.py | applications/ColossalChat/coati/trainer/__init__.py | from .base import OLTrainer, SLTrainer
from .dpo import DPOTrainer
from .grpo import GRPOTrainer
from .kto import KTOTrainer
from .orpo import ORPOTrainer
from .ppo import PPOTrainer
from .rm import RewardModelTrainer
from .sft import SFTTrainer
__all__ = [
"SLTrainer",
"OLTrainer",
"RewardModelTrainer",
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/grpo.py | applications/ColossalChat/coati/trainer/grpo.py | """
GRPO trainer
"""
import os
from typing import Dict, List, Optional, Union
import torch
import wandb
from coati.experience_buffer import NaiveExperienceBuffer
from coati.experience_maker import Experience, NaiveExperienceMaker
from coati.models import RewardModel, RLVRRewardModel
from coati.models.loss import GPTL... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/base.py | applications/ColossalChat/coati/trainer/base.py | """
Base trainers for online and offline training
SLTrainer: supervised learning trainer
pretrain, sft, dpo, reward model training
OLTrainer: online learning trainer
rlhf-ppo
"""
from abc import ABC, abstractmethod
from contextlib import contextmanager
from typing import Callable, List
import ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/dpo.py | applications/ColossalChat/coati/trainer/dpo.py | """
Dpo trainer
"""
import os
from typing import Any, Optional
import torch
import torch.distributed as dist
from coati.models.loss import DpoLoss
from coati.models.utils import calc_masked_log_probs
from coati.trainer.utils import all_reduce_mean
from coati.utils import AccumulativeMeanMeter, save_checkpoint
from to... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/callbacks/__init__.py | applications/ColossalChat/coati/trainer/callbacks/__init__.py | from .base import Callback
from .performance_evaluator import PerformanceEvaluator
__all__ = ["Callback", "PerformanceEvaluator"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/callbacks/base.py | applications/ColossalChat/coati/trainer/callbacks/base.py | from abc import ABC
from coati.experience_maker import Experience
class Callback(ABC):
"""
Base callback class. It defines the interface for callbacks.
"""
def on_fit_start(self) -> None:
pass
def on_fit_end(self) -> None:
pass
def on_episode_start(self, episode: int) -> No... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/trainer/callbacks/performance_evaluator.py | applications/ColossalChat/coati/trainer/callbacks/performance_evaluator.py | from time import time
from typing import Optional
import torch
import torch.distributed as dist
from coati.experience_maker import Experience
from .base import Callback
def get_world_size() -> int:
if dist.is_initialized():
return dist.get_world_size()
return 1
def save_eval_result_rank_0(s: str, ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/accumulative_meter.py | applications/ColossalChat/coati/utils/accumulative_meter.py | """
A class that can be used to calculate the mean of a variable
"""
class AccumulativeMeanVariable:
"""
A class that calculates the accumulative mean of a variable.
"""
def __init__(self):
self._sum = 0
self._count = 0
def add(self, value, count_update=1):
"""
Ad... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/ckpt_io.py | applications/ColossalChat/coati/utils/ckpt_io.py | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Helper functions for IO save load checkpoints
"""
import json
import os
from typing import Any, Dict, Tuple, Union
import torch
from torch.optim.lr_scheduler import _LRScheduler
from torch.optim.optimizer import Optimizer
from colossalai.booster import Booster
from... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/__init__.py | applications/ColossalChat/coati/utils/__init__.py | from .accumulative_meter import AccumulativeMeanMeter
from .ckpt_io import load_checkpoint, save_checkpoint
__all__ = ["load_checkpoint", "save_checkpoint", "AccumulativeMeanMeter"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/reward_score/gsm8k.py | applications/ColossalChat/coati/utils/reward_score/gsm8k.py | import torch
from .utils import extract_solution, validate_response_structure
def gsm8k_reward_fn(input_ids, attention_mask, **kwargs):
# apply varifiable reward
# reward 10 points if the final answer is correct, reward 1 point if format is correct
gt_answer = kwargs["gt_answer"]
tokenizer = kwargs[... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/reward_score/utils.py | applications/ColossalChat/coati/utils/reward_score/utils.py | # Copyright Unakar
# Modified from https://github.com/Unakar/Logic-RL/blob/086373176ac198c97277ff50f4b6e7e1bfe669d3/verl/utils/reward_score/kk.py#L99
#
# 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 Lice... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/reward_score/__init__.py | applications/ColossalChat/coati/utils/reward_score/__init__.py | from .competition import math_competition_reward_fn
from .gsm8k import gsm8k_reward_fn
__all__ = ["gsm8k_reward_fn", "math_competition_reward_fn"]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/utils/reward_score/competition.py | applications/ColossalChat/coati/utils/reward_score/competition.py | import torch
from .utils import extract_solution, validate_response_structure
def math_competition_reward_fn(input_ids, attention_mask, **kwargs):
# apply varifiable reward
# reward 10 points if the final answer is correct, reward 1 point if format is correct
gt_answer = kwargs["gt_answer"]
tokenize... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/dataset/loader.py | applications/ColossalChat/coati/dataset/loader.py | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Dataloader for sft, dpo, ppo
"""
import os
from dataclasses import dataclass
from typing import Dict, Iterator, List, Optional, Sequence, Union
import jsonlines
import torch
import torch.nn.functional as F
from coati.dataset.utils import chuncate_sequence, pad_to_max... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/dataset/conversation.py | applications/ColossalChat/coati/dataset/conversation.py | import dataclasses
import json
import os
from typing import Any, Dict, List
import torch.distributed as dist
from transformers import AutoTokenizer, PreTrainedTokenizer
from colossalai.logging import get_dist_logger
logger = get_dist_logger()
@dataclasses.dataclass
class Conversation:
tokenizer: PreTrainedToke... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/dataset/utils.py | applications/ColossalChat/coati/dataset/utils.py | import io
import json
from typing import Any, Dict, List
import torch
import torch.distributed as dist
import torch.nn.functional as F
from transformers import PreTrainedTokenizer
def is_rank_0() -> bool:
return not dist.is_initialized() or dist.get_rank() == 0
def _make_r_io_base(f, mode: str):
if not isi... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/dataset/tokenization_utils.py | applications/ColossalChat/coati/dataset/tokenization_utils.py | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
tokenization utils for constructing dataset for ppo, dpo, sft, rm
"""
import warnings
from copy import deepcopy
from typing import Any, Dict, List, Union
from coati.dataset.conversation import Conversation
from coati.dataset.utils import split_templated_prompt_into_c... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/dataset/__init__.py | applications/ColossalChat/coati/dataset/__init__.py | from .conversation import Conversation, setup_conversation_template
from .loader import (
DataCollatorForKTODataset,
DataCollatorForPreferenceDataset,
DataCollatorForPromptDataset,
DataCollatorForSupervisedDataset,
StatefulDistributedSampler,
load_tokenized_dataset,
)
from .tokenization_utils im... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/detached_trainer_base.py | applications/ColossalChat/coati/ray/detached_trainer_base.py | import os
from abc import ABC, abstractmethod
from typing import Any, Dict, List
import ray
import torch
from coati.experience_buffer.utils import BufferItem
from coati.experience_maker import Experience
from torch.utils.data import DataLoader
from tqdm import tqdm
from .callbacks import TrainerCallback
from .detache... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/utils.py | applications/ColossalChat/coati/ray/utils.py | import os
from collections import OrderedDict
from typing import Any, Dict
import torch
import torch.distributed as dist
import torch.nn as nn
from coati.models.bloom import BLOOMRM, BLOOMActor, BLOOMCritic
from coati.models.gpt import GPTRM, GPTActor, GPTCritic
from coati.models.llama import LlamaActor, LlamaCritic, ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/detached_trainer_ppo.py | applications/ColossalChat/coati/ray/detached_trainer_ppo.py | from typing import Callable, Dict, List, Tuple
import ray
import torch
from coati.experience_maker import Experience
from coati.models.base import Actor, Critic
from coati.models.loss import PolicyLoss, ValueLoss
from coati.trainer.strategies import GeminiStrategy, LowLevelZeroStrategy, Strategy
from torch.optim impor... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/experience_maker_holder.py | applications/ColossalChat/coati/ray/experience_maker_holder.py | import os
import time
import tracemalloc
from threading import Lock
from typing import Any, Callable, Dict, Iterable, List, Tuple, Union
import ray
import torch
from coati.experience_buffer.utils import split_experience_batch
from coati.experience_maker import Experience, NaiveExperienceMaker
from coati.models.base im... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/__init__.py | applications/ColossalChat/coati/ray/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/lora_constructor.py | applications/ColossalChat/coati/ray/lora_constructor.py | from collections import OrderedDict
from dataclasses import dataclass
from typing import Any, Dict
import torch.nn as nn
from coati.models.lora import LoraLinear
@dataclass
class LoRAConfig:
r: int = 0
lora_alpha: int = 1
lora_dropout: float = 0
fan_in_fan_out: bool = False
class LoRAConstructor:
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/detached_replay_buffer.py | applications/ColossalChat/coati/ray/detached_replay_buffer.py | from typing import List
import torch
from coati.experience_buffer.utils import BufferItem, make_experience_batch, split_experience_batch
from coati.experience_maker.base import Experience
# from torch.multiprocessing import Queue
from ray.util.queue import Queue
class DetachedReplayBuffer:
"""
Detached ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/callbacks/__init__.py | applications/ColossalChat/coati/ray/callbacks/__init__.py | from .base import MakerCallback, TrainerCallback
from .performance_evaluator import ExperienceMakerPerformanceEvaluator, TrainerPerformanceEvaluator
__all__ = [
"TrainerCallback",
"MakerCallback",
"ExperienceMakerPerformanceEvaluator",
"TrainerPerformanceEvaluator",
]
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/callbacks/base.py | applications/ColossalChat/coati/ray/callbacks/base.py | from abc import ABC
from coati.experience_maker import Experience
class TrainerCallback(ABC):
"""
Base callback class. It defines the interface for callbacks.
"""
def on_fit_start(self) -> None:
pass
def on_fit_end(self) -> None:
pass
def on_episode_start(self, episode: int... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/coati/ray/callbacks/performance_evaluator.py | applications/ColossalChat/coati/ray/callbacks/performance_evaluator.py | from time import time
from typing import Optional
import torch
import torch.distributed as dist
from coati.experience_maker import Experience
from .base import MakerCallback, TrainerCallback
def get_world_size() -> int:
if dist.is_initialized():
return dist.get_world_size()
return 1
def print_rank... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/benchmarks/dummy_dataset.py | applications/ColossalChat/benchmarks/dummy_dataset.py | from typing import Callable
from torch.utils.data import Dataset
class DummyLLMDataset(Dataset):
def __init__(self, keys, seq_len, size=500, gen_fn={}):
self.keys = keys
self.gen_fn = gen_fn
self.seq_len = seq_len
self.data = self._generate_data()
self.size = size
def... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/benchmarks/prepare_dummy_test_dataset.py | applications/ColossalChat/benchmarks/prepare_dummy_test_dataset.py | import argparse
import json
import os
import time
from multiprocessing import cpu_count
from datasets import load_dataset
from dummy_dataset import DummyLLMDataset
from colossalai.logging import get_dist_logger
logger = get_dist_logger()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/benchmarks/benchmark_ppo.py | applications/ColossalChat/benchmarks/benchmark_ppo.py | """
For becnhmarking ppo. Mudified from examples/training_scripts/train_ppo.py
"""
import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
import torch.distributed as dist
from coati.dataset import (
DataCollatorForPromptDataset,
DataCollatorForSupervisedDataset,
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/benchmarks/ray/1mmt_dummy.py | applications/ColossalChat/benchmarks/ray/1mmt_dummy.py | import argparse
import os
import socket
from functools import partial
import ray
import torch
from coati.quant import llama_load_quant, low_resource_init
from coati.ray.detached_trainer_ppo import DetachedPPOTrainer
from coati.ray.experience_maker_holder import ExperienceMakerHolder
from coati.ray.utils import (
g... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/benchmarks/ray/mmmt_dummy.py | applications/ColossalChat/benchmarks/ray/mmmt_dummy.py | import argparse
import os
import socket
from functools import partial
import ray
import torch
from coati.quant import llama_load_quant, low_resource_init
from coati.ray.detached_trainer_ppo import DetachedPPOTrainer
from coati.ray.experience_maker_holder import ExperienceMakerHolder
from coati.ray.utils import (
g... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/tests/verify_chat_data.py | applications/ColossalChat/tests/verify_chat_data.py | import argparse
import json
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--data_source",
type=str,
required=True,
default=None,
help="The raw data file",
)
parser.add_argument(
"--to_verify_file",
type=str,
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/tests/generate_dummy_datasets_for_testing.py | applications/ColossalChat/tests/generate_dummy_datasets_for_testing.py | import argparse
import json
import os
sft_seed = {
"messages": [
{"from": "user", "content": "Give three tips for staying healthy."},
{
"from": "assistant",
"content": "1.Eat a balanced diet and make sure to include plenty of fruits and vegetables. \n2. Exercise regularly to... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/tests/__init__.py | applications/ColossalChat/tests/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/tests/test_lora.py | applications/ColossalChat/tests/test_lora.py | import torch
import torch.nn as nn
import torch.optim as optim
from coati.models import convert_to_lora_module
from coati.models.lora import LoraConfig, LoraEmbedding, LoraLinear
from torch.utils.data import DataLoader, TensorDataset
class SimpleNN(nn.Module):
def __init__(self, input_size, hidden_size, num_class... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/ray/train_prompts_on_ray.py | applications/ColossalChat/examples/community/ray/train_prompts_on_ray.py | import argparse
import logging
import os
import socket
from copy import deepcopy
from typing import Type
import ray
import torch
from coati.experience_maker.base import Experience
from coati.models.base import RewardModel
from coati.models.bloom import BLOOMActor, BLOOMCritic
from coati.models.gpt import GPTActor, GPT... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/ray/ray_job_script.py | applications/ColossalChat/examples/community/ray/ray_job_script.py | import sys
from ray.job_submission import JobSubmissionClient
def main(api_server_endpoint="http://127.0.0.1:8265"):
client = JobSubmissionClient(api_server_endpoint)
client.submit_job(
entrypoint="python experimental/ray/train_prompts_on_ray.py --strategy colossalai_zero2 --prompt_csv_url https://hu... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/peft/train_peft_sft.py | applications/ColossalChat/examples/community/peft/train_peft_sft.py | import argparse
import os
import torch
import torch.distributed as dist
from coati.trainer import SFTTrainer
from coati.trainer.strategies import DDPStrategy, GeminiStrategy, LowLevelZeroStrategy
from easy_dataset import EasyDataset
from peft import LoraConfig, PeftModel, TaskType, get_peft_model
from torch.optim impo... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/peft/easy_dataset.py | applications/ColossalChat/examples/community/peft/easy_dataset.py | import copy
import json
from typing import Dict, Sequence
import torch
from torch.utils.data import Dataset
from tqdm import tqdm
from transformers import AutoTokenizer
IGNORE_INDEX = -100
def _tokenize_fn(strings: Sequence[str], tokenizer: AutoTokenizer, max_length: int = 512) -> Dict:
"""Tokenize a list of st... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/peft/train_peft_prompts.py | applications/ColossalChat/examples/community/peft/train_peft_prompts.py | import argparse
import torch
import torch.distributed as dist
from coati.dataset import DataCollatorForSupervisedDataset
from coati.models.bloom import BLOOMRM, BLOOMCritic
from coati.models.gpt import GPTRM, GPTCritic
from coati.models.llama import LlamaCritic, LlamaRM
from coati.models.opt import OPTRM, OPTCritic
fr... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/community/peft/easy_models.py | applications/ColossalChat/examples/community/peft/easy_models.py | from typing import Optional, Tuple, Union
import torch
import torch.nn as nn
import torch.nn.functional as F
from coati.models.generation import generate
from coati.models.utils import log_probs_from_logits
from peft import PeftModel
from torch.nn.modules import Module
from transformers import BloomConfig, BloomForCau... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_ppo.py | applications/ColossalChat/examples/training_scripts/train_ppo.py | import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
import torch.distributed as dist
from coati.dataset import (
DataCollatorForPromptDataset,
DataCollatorForSupervisedDataset,
StatefulDistributedSampler,
load_tokenized_dataset,
setup_conversation_t... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_orpo.py | applications/ColossalChat/examples/training_scripts/train_orpo.py | import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
from coati.dataset import DataCollatorForPreferenceDataset, StatefulDistributedSampler, load_tokenized_dataset
from coati.models import LoraConfig, convert_to_lora_module, disable_dropout
from coati.trainer import ORP... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_kto.py | applications/ColossalChat/examples/training_scripts/train_kto.py | import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
from coati.dataset import DataCollatorForKTODataset, StatefulDistributedSampler, load_tokenized_dataset
from coati.models import LoraConfig, convert_to_lora_module, disable_dropout
from coati.trainer import KTOTrainer... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_sft.py | applications/ColossalChat/examples/training_scripts/train_sft.py | import argparse
import json
import math
import os
import resource
from contextlib import nullcontext
import torch
from coati.dataset import DataCollatorForSupervisedDataset, StatefulDistributedSampler, load_tokenized_dataset
from coati.models import LoraConfig, convert_to_lora_module
from coati.trainer import SFTTrain... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_grpo.py | applications/ColossalChat/examples/training_scripts/train_grpo.py | import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
import torch.distributed as dist
from coati.dataset import (
DataCollatorForPromptDataset,
DataCollatorForSupervisedDataset,
StatefulDistributedSampler,
load_tokenized_dataset,
setup_conversation_t... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_dpo.py | applications/ColossalChat/examples/training_scripts/train_dpo.py | import argparse
import json
import os
import resource
from contextlib import nullcontext
import torch
from coati.dataset import DataCollatorForPreferenceDataset, StatefulDistributedSampler, load_tokenized_dataset
from coati.models import LoraConfig, convert_to_lora_module, disable_dropout
from coati.trainer import DPO... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/lora_finetune.py | applications/ColossalChat/examples/training_scripts/lora_finetune.py | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Supervised fine-tuning of MoE models like Deepseek V3/R1 on a downstream task.
"""
import argparse
import json
import os
import resource
from contextlib import nullcontext
from types import MethodType
import torch
import torch.distributed as dist
from coati.dataset.l... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/training_scripts/train_rm.py | applications/ColossalChat/examples/training_scripts/train_rm.py | import argparse
import json
import math
import os
import resource
from contextlib import nullcontext
import torch
from coati.dataset import DataCollatorForPreferenceDataset, StatefulDistributedSampler, load_tokenized_dataset
from coati.models import LogExpLoss, LogSigLoss, LoraConfig, RewardModel, convert_to_lora_modu... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/inference/inference.py | applications/ColossalChat/examples/inference/inference.py | import argparse
import json
import os
from typing import Dict
import torch
from chatio import dummy_io, rich_io, simple_io
from coati.dataset.conversation import setup_conversation_template
from coati.models import generate_streaming
from transformers import AutoModelForCausalLM, AutoTokenizer, PreTrainedModel
from c... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/inference/chatio.py | applications/ColossalChat/examples/inference/chatio.py | """
command line IO utils for chatbot
"""
import abc
import re
from prompt_toolkit import PromptSession
from prompt_toolkit.auto_suggest import AutoSuggestFromHistory
from prompt_toolkit.completion import WordCompleter
from prompt_toolkit.history import InMemoryHistory
from rich.console import Console
from rich.live ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/inference/web_chatbot/utils.py | applications/ColossalChat/examples/inference/web_chatbot/utils.py | import copy
import json
from threading import Lock
from typing import List
import jieba
import torch
from coati.dataset.conversation import default_conversation
from pydantic import BaseModel, Field
def update_model_kwargs_fn(outputs: dict, **model_kwargs) -> dict:
if "past_key_values" in outputs:
model_... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/inference/web_chatbot/locustfile.py | applications/ColossalChat/examples/inference/web_chatbot/locustfile.py | from locust import HttpUser, task
samples = [
[
dict(
instruction="Who is the best player in the history of NBA?",
response="The best player in the history of the NBA is widely considered to be Michael Jordan. He is one of the most successful players in the league, having won 6 NBA ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/inference/web_chatbot/server.py | applications/ColossalChat/examples/inference/web_chatbot/server.py | import argparse
import os
from threading import Lock
from typing import Generator, List, Optional
import torch
import uvicorn
from coati.models import generate_streaming
from coati.quant import llama_load_quant, low_resource_init
from fastapi import FastAPI, Request
from fastapi.middleware.cors import CORSMiddleware
f... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/applications/ColossalChat/examples/data_preparation_scripts/prepare_dataset.py | applications/ColossalChat/examples/data_preparation_scripts/prepare_dataset.py | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Prepare dataset scripts
Usage:
- For SFT dataset preparation (SFT)
python prepare_dataset.py --type sft \
--data_input_dirs /PATH/TO/SFT/DATASET \
--conversation_template_config /PATH/TO/CHAT/TEMPLATE/CONFIG.json \
--tokenizer_dir "" \
--data_cache_di... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/__init__.py | colossalai/__init__.py | from . import accelerator
from .initialize import launch, launch_from_openmpi, launch_from_slurm, launch_from_torch
try:
# .version will be created by setup.py
from .version import __version__
except ModuleNotFoundError:
# this will only happen if the user did not run `pip install`
# and directly set P... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/initialize.py | colossalai/initialize.py | #!/usr/bin/env python
# -*- encoding: utf-8 -*-
import os
# set CUDA_DEVICE_MAX_CONNECTIONS=1 to ensure that when overlapping communication and computation,
# the order of of kernel launches on GPUs are the same as on the CPU so that comm is launched first.
# see https://github.com/NVIDIA/Megatron-LM/issues/533
# htt... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/__init__.py | colossalai/_analyzer/__init__.py | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false | |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/envs.py | colossalai/_analyzer/envs.py | from dataclasses import dataclass
@dataclass
class MeshConfig:
TFLOPS: float = 1.9e12
BANDWIDTH = 1.2e9
| python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/_subclasses/meta_tensor.py | colossalai/_analyzer/_subclasses/meta_tensor.py | import uuid
from functools import partial
import torch
import torch.distributed as dist
from torch.types import _device
from torch.utils._pytree import tree_map
from ._monkey_patch import _AliasATen, _DistCommMethod, _InplaceATen, _MaybeInplaceATen, _TorchOverrideableFactoryMethod
__all__ = ["MetaTensor", "MetaTenso... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/_subclasses/_monkey_patch.py | colossalai/_analyzer/_subclasses/_monkey_patch.py | import torch
from packaging import version
__all__ = [
"_TorchFactoryMethod",
"_TorchOverrideableFactoryMethod",
"_TorchNonOverrideableFactoryMethod",
"_TensorPropertyMethod",
"_DistCommMethod",
"_AliasATen",
"_InplaceATen",
"_MaybeInplaceATen",
]
_TorchOverrideableFactoryMethod = [
... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/_subclasses/_meta_registration.py | colossalai/_analyzer/_subclasses/_meta_registration.py | # meta patch from https://github.com/pytorch/pytorch/blob/master/torch/_meta_registrations.py
# should be activated for PyTorch version 1.12.0 and below
# refer to https://github.com/pytorch/pytorch/blob/master/aten/src/ATen/native/native_functions.yaml
# for more meta_registrations
from typing import List, Optional, ... | python | Apache-2.0 | b1915d2889543949eb5b610241f1515c73df5059 | 2026-01-04T14:40:19.002665Z | false |
hpcaitech/ColossalAI | https://github.com/hpcaitech/ColossalAI/blob/b1915d2889543949eb5b610241f1515c73df5059/colossalai/_analyzer/_subclasses/__init__.py | colossalai/_analyzer/_subclasses/__init__.py | from ._meta_registration import *
from ._monkey_patch import *
from .flop_tensor import flop_count, flop_mapping
from .meta_tensor import MetaTensor, MetaTensorMode
| 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.