2021-10-28 16:21:23 +00:00
|
|
|
#!/usr/bin/env python
|
|
|
|
# -*- encoding: utf-8 -*-
|
2022-11-17 05:42:33 +00:00
|
|
|
import functools
|
2022-04-01 01:46:33 +00:00
|
|
|
import os
|
2021-12-29 15:32:10 +00:00
|
|
|
import random
|
|
|
|
import socket
|
2022-12-23 12:57:41 +00:00
|
|
|
from collections import defaultdict
|
|
|
|
from contextlib import contextmanager
|
2022-04-01 01:46:33 +00:00
|
|
|
from pathlib import Path
|
2022-11-17 05:42:33 +00:00
|
|
|
from typing import Callable, Dict, List, Optional, Union
|
2022-08-11 14:58:58 +00:00
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
import torch
|
2022-12-23 12:57:41 +00:00
|
|
|
import torch.distributed as dist
|
2023-02-17 01:22:45 +00:00
|
|
|
from torch import inf
|
2022-03-01 10:17:01 +00:00
|
|
|
from torch.nn.parameter import Parameter
|
|
|
|
|
2022-11-17 05:42:33 +00:00
|
|
|
from colossalai.constants import IS_TENSOR_PARALLEL, NUM_PARTITIONS, TENSOR_PARALLEL_ATTRIBUTES
|
2021-10-28 16:21:23 +00:00
|
|
|
from colossalai.context.parallel_mode import ParallelMode
|
|
|
|
from colossalai.core import global_context as gpc
|
2022-02-14 03:15:02 +00:00
|
|
|
from colossalai.global_variables import tensor_parallel_env as env
|
2022-08-11 14:58:58 +00:00
|
|
|
from colossalai.tensor import ColoParameter, ProcessGroup
|
2022-11-17 05:42:33 +00:00
|
|
|
|
|
|
|
from .multi_tensor_apply import multi_tensor_applier
|
2022-08-11 14:58:58 +00:00
|
|
|
|
2023-01-06 12:50:26 +00:00
|
|
|
try:
|
|
|
|
from colossalai._C import fused_optim
|
|
|
|
except:
|
|
|
|
fused_optim = None
|
|
|
|
|
2021-10-28 16:21:23 +00:00
|
|
|
|
|
|
|
def print_rank_0(msg: str, logger=None):
|
2022-01-21 02:44:30 +00:00
|
|
|
"""Print messages and save logs(optional). This is executed only if you are the rank-0 gpu.
|
2021-10-28 16:21:23 +00:00
|
|
|
|
2022-03-25 05:02:39 +00:00
|
|
|
Args:
|
|
|
|
msg (str): A string message to output.
|
|
|
|
logger (:class:`colossalai.logging.DistributedLogger`, optional):
|
|
|
|
The logger to record the message, defaults to None.
|
2022-01-21 02:44:30 +00:00
|
|
|
"""
|
2021-10-28 16:21:23 +00:00
|
|
|
if gpc.get_global_rank() == 0:
|
|
|
|
if logger is None:
|
|
|
|
print(msg, flush=True)
|
|
|
|
else:
|
|
|
|
logger.info(msg)
|
|
|
|
|
|
|
|
|
2022-04-01 01:46:33 +00:00
|
|
|
def ensure_path_exists(filename: str):
|
|
|
|
# ensure the path exists
|
|
|
|
dirpath = os.path.dirname(filename)
|
|
|
|
if not os.path.exists(dirpath):
|
|
|
|
Path(dirpath).mkdir(parents=True, exist_ok=True)
|
|
|
|
|
|
|
|
|
2021-12-29 15:32:10 +00:00
|
|
|
def free_port():
|
|
|
|
while True:
|
|
|
|
try:
|
|
|
|
sock = socket.socket()
|
2022-03-17 06:40:52 +00:00
|
|
|
sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
|
2021-12-29 15:32:10 +00:00
|
|
|
port = random.randint(20000, 65000)
|
|
|
|
sock.bind(('localhost', port))
|
|
|
|
sock.close()
|
|
|
|
return port
|
|
|
|
except Exception:
|
|
|
|
continue
|
|
|
|
|
|
|
|
|
2022-01-20 05:44:51 +00:00
|
|
|
def sync_model_param(model, parallel_mode):
|
2022-03-25 05:02:39 +00:00
|
|
|
r"""Make sure data parameters are consistent during Data Parallel Mode.
|
2021-10-28 16:21:23 +00:00
|
|
|
|
2022-03-25 05:02:39 +00:00
|
|
|
Args:
|
|
|
|
model (:class:`torch.nn.Module`): A pyTorch model on whose parameters you check the consistency.
|
|
|
|
parallel_mode (:class:`colossalai.context.ParallelMode`): Parallel mode to be checked.
|
|
|
|
|
|
|
|
Note:
|
|
|
|
The parallel_mode should be concluded in ``ParallelMode``. More details about ``ParallelMode`` could be found
|
|
|
|
in `parallel_mode <https://github.com/hpcaitech/ColossalAI/blob/main/colossalai/context/parallel_mode.py>`_
|
2022-01-21 02:44:30 +00:00
|
|
|
"""
|
2022-01-20 05:44:51 +00:00
|
|
|
if gpc.is_initialized(parallel_mode) and gpc.get_world_size(parallel_mode) > 1:
|
2021-10-28 16:21:23 +00:00
|
|
|
for param in model.parameters():
|
2022-01-20 05:44:51 +00:00
|
|
|
ranks = gpc.get_ranks_in_group(parallel_mode)
|
2022-02-14 03:15:02 +00:00
|
|
|
dist.broadcast(param, src=ranks[0], group=gpc.get_group(parallel_mode))
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
|
2021-10-28 16:21:23 +00:00
|
|
|
|
|
|
|
def is_dp_rank_0():
|
|
|
|
return not gpc.is_initialized(ParallelMode.DATA) or gpc.is_first_rank(ParallelMode.DATA)
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
|
2021-10-28 16:21:23 +00:00
|
|
|
def is_tp_rank_0():
|
|
|
|
return not gpc.is_initialized(ParallelMode.TENSOR) or gpc.is_first_rank(ParallelMode.TENSOR)
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
|
2021-10-28 16:21:23 +00:00
|
|
|
def is_no_pp_or_last_stage():
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
return not gpc.is_initialized(ParallelMode.PIPELINE) or gpc.is_last_rank(ParallelMode.PIPELINE)
|
|
|
|
|
|
|
|
|
|
|
|
def is_using_ddp():
|
|
|
|
return gpc.is_initialized(ParallelMode.DATA) and gpc.get_world_size(ParallelMode.DATA) > 1
|
|
|
|
|
|
|
|
|
|
|
|
def is_using_pp():
|
|
|
|
return gpc.is_initialized(ParallelMode.PIPELINE) and gpc.get_world_size(ParallelMode.PIPELINE) > 1
|
|
|
|
|
|
|
|
|
2022-01-20 05:44:51 +00:00
|
|
|
def is_using_sequence():
|
|
|
|
return gpc.is_initialized(ParallelMode.SEQUENCE) and gpc.get_world_size(ParallelMode.SEQUENCE) > 1
|
|
|
|
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
@contextmanager
|
|
|
|
def conditional_context(context_manager, enable=True):
|
|
|
|
if enable:
|
|
|
|
with context_manager:
|
|
|
|
yield
|
|
|
|
else:
|
|
|
|
yield
|
|
|
|
|
|
|
|
|
2022-02-14 03:15:02 +00:00
|
|
|
class model_branch_context(object):
|
2022-04-11 02:46:08 +00:00
|
|
|
|
2022-02-14 03:15:02 +00:00
|
|
|
def __enter__(self):
|
|
|
|
self.env_status = env.save()
|
|
|
|
|
|
|
|
def __exit__(self, *exc_info):
|
|
|
|
env.load(**self.env_status)
|
|
|
|
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
def is_model_parallel_parameter(p):
|
|
|
|
return hasattr(p, IS_TENSOR_PARALLEL) and getattr(p, IS_TENSOR_PARALLEL)
|
|
|
|
|
|
|
|
|
2023-01-11 04:22:45 +00:00
|
|
|
def is_ddp_ignored(p):
|
|
|
|
return getattr(p, '_ddp_to_ignore', False)
|
|
|
|
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
def _calc_l2_norm(grads):
|
2023-01-11 04:22:45 +00:00
|
|
|
# we should not
|
2023-01-06 12:50:26 +00:00
|
|
|
global fused_optim
|
|
|
|
|
|
|
|
if fused_optim is None:
|
|
|
|
from colossalai.kernel.op_builder import FusedOptimBuilder
|
|
|
|
fused_optim = FusedOptimBuilder().load()
|
2023-01-11 04:22:45 +00:00
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
norm = 0.0
|
|
|
|
if len(grads) > 0:
|
|
|
|
dummy_overflow_buf = torch.cuda.IntTensor([0])
|
|
|
|
norm, _ = multi_tensor_applier(
|
2022-12-23 09:07:03 +00:00
|
|
|
fused_optim.multi_tensor_l2norm,
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
dummy_overflow_buf,
|
|
|
|
[grads],
|
2022-04-11 02:46:08 +00:00
|
|
|
False # no per-parameter norm
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
)
|
|
|
|
return norm
|
|
|
|
|
|
|
|
|
|
|
|
def _calc_lp(grads, norm_type):
|
|
|
|
norm = 0.0
|
|
|
|
for grad in grads:
|
|
|
|
grad_norm = torch.norm(grad, norm_type)
|
2022-02-14 03:15:02 +00:00
|
|
|
norm += grad_norm**norm_type
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
return norm
|
|
|
|
|
2022-03-21 05:35:04 +00:00
|
|
|
|
2022-03-01 10:17:01 +00:00
|
|
|
def _move_norm_to_cuda(norm: Union[float, torch.Tensor]) -> Union[float, torch.Tensor]:
|
|
|
|
if torch.is_tensor(norm) and norm.device.type != 'cuda':
|
|
|
|
norm = norm.to(torch.cuda.current_device())
|
|
|
|
return norm
|
2022-02-14 03:15:02 +00:00
|
|
|
|
2022-03-21 05:35:04 +00:00
|
|
|
|
2022-06-27 01:53:57 +00:00
|
|
|
def _get_tensor_norm(norm: Union[float, torch.Tensor], move_to_cuda) -> torch.Tensor:
|
|
|
|
if isinstance(norm, float):
|
|
|
|
norm = torch.Tensor([norm])
|
|
|
|
if move_to_cuda:
|
|
|
|
norm = norm.to(torch.cuda.current_device())
|
|
|
|
return norm
|
|
|
|
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
# ======== Gradient Clipping =========
|
|
|
|
|
|
|
|
|
2022-08-11 14:58:58 +00:00
|
|
|
def _compute_local_lp(params: List[ColoParameter], norm_type: float) -> float:
|
|
|
|
if len(params) == 0:
|
|
|
|
return 0.0
|
|
|
|
grads = [p.grad for p in params]
|
|
|
|
use_cuda_kernel = grads[0].device.type == 'cuda'
|
|
|
|
if norm_type == inf:
|
|
|
|
local_lp = max([g.abs().max() for g in grads])
|
|
|
|
elif norm_type == 2.0 and use_cuda_kernel:
|
|
|
|
local_lp = _calc_l2_norm(grads)**norm_type
|
|
|
|
else:
|
|
|
|
local_lp = _calc_lp(grads, norm_type)
|
|
|
|
if isinstance(local_lp, torch.Tensor):
|
|
|
|
return local_lp.item()
|
|
|
|
return local_lp
|
|
|
|
|
|
|
|
|
|
|
|
def _compute_buckets_lp(params: List[ColoParameter], norm_type: float) -> float:
|
|
|
|
if len(params) == 0:
|
|
|
|
return 0.0
|
|
|
|
buckets: Dict[Optional[ProcessGroup], List[ColoParameter]] = defaultdict(list)
|
|
|
|
for p in params:
|
|
|
|
if p.is_replicate():
|
|
|
|
buckets[None].append(p)
|
|
|
|
else:
|
|
|
|
buckets[p.get_process_group().tp_process_group()].append(p)
|
|
|
|
total_lp = 0.0
|
|
|
|
for group, bucket in buckets.items():
|
|
|
|
local_lp = _compute_local_lp(bucket, norm_type)
|
|
|
|
if group is not None:
|
|
|
|
local_lp_tensor = torch.tensor([local_lp], device=torch.cuda.current_device())
|
|
|
|
if norm_type == inf:
|
|
|
|
dist.all_reduce(local_lp_tensor, op=dist.ReduceOp.MAX, group=group)
|
|
|
|
else:
|
|
|
|
dist.all_reduce(local_lp_tensor, group=group)
|
|
|
|
local_lp = local_lp_tensor.item()
|
|
|
|
if norm_type == inf:
|
|
|
|
total_lp = max(total_lp, local_lp)
|
|
|
|
else:
|
|
|
|
total_lp += local_lp
|
|
|
|
return total_lp
|
|
|
|
|
|
|
|
|
|
|
|
def _compute_pp_grad_lp(total_lp: float, norm_type: float) -> float:
|
|
|
|
if gpc.is_initialized(ParallelMode.PIPELINE) and gpc.get_world_size(ParallelMode.PIPELINE) > 1:
|
|
|
|
total_lp_tensor = torch.tensor([total_lp], device=torch.cuda.current_device())
|
|
|
|
if norm_type == inf:
|
|
|
|
dist.all_reduce(total_lp_tensor, op=dist.ReduceOp.MAX, group=gpc.get_group(ParallelMode.PIPELINE))
|
|
|
|
else:
|
|
|
|
dist.all_reduce(total_lp_tensor, group=gpc.get_group(ParallelMode.PIPELINE))
|
|
|
|
total_lp = total_lp_tensor.item()
|
|
|
|
return total_lp
|
|
|
|
|
|
|
|
|
|
|
|
def _compute_grad_lp(parameters, norm_type: float = 2.0) -> float:
|
|
|
|
if isinstance(parameters, torch.Tensor):
|
|
|
|
parameters = [parameters]
|
|
|
|
grad_dtype = None
|
|
|
|
cpu_grad_params: List[ColoParameter] = []
|
|
|
|
cuda_grad_params: List[ColoParameter] = []
|
|
|
|
for p in parameters:
|
|
|
|
if p.grad is None:
|
|
|
|
continue
|
|
|
|
assert isinstance(p, ColoParameter)
|
|
|
|
if grad_dtype is None:
|
|
|
|
grad_dtype = p.grad.dtype
|
|
|
|
assert p.grad.dtype == grad_dtype, f'Expected all grads are {grad_dtype}, got {p.grad.dtype}'
|
|
|
|
if p.grad.device.type == 'cuda':
|
|
|
|
cuda_grad_params.append(p)
|
|
|
|
else:
|
|
|
|
cpu_grad_params.append(p)
|
|
|
|
norm_type = float(norm_type)
|
|
|
|
cpu_lp = _compute_buckets_lp(cpu_grad_params, norm_type)
|
|
|
|
cuda_lp = _compute_buckets_lp(cuda_grad_params, norm_type)
|
|
|
|
if norm_type == inf:
|
|
|
|
total_lp = max(cpu_lp, cuda_lp)
|
|
|
|
else:
|
|
|
|
total_lp = cpu_lp + cuda_lp
|
|
|
|
return _compute_pp_grad_lp(total_lp, norm_type)
|
|
|
|
|
|
|
|
|
|
|
|
def compute_grad_norm(parameters, norm_type: float = 2.0) -> float:
|
|
|
|
norm_type = float(norm_type)
|
|
|
|
total_norm = _compute_grad_lp(parameters, norm_type)
|
|
|
|
if norm_type != inf:
|
|
|
|
total_norm = total_norm**(1 / norm_type)
|
|
|
|
return total_norm
|
|
|
|
|
|
|
|
|
|
|
|
def _clip_grad_norm(parameters, max_norm: float, total_norm: float) -> None:
|
|
|
|
clip_coef = max_norm / (total_norm + 1e-6)
|
|
|
|
if clip_coef < 1.0:
|
|
|
|
cuda_grads: List[torch.Tensor] = []
|
|
|
|
cpu_grads: List[torch.Tensor] = []
|
|
|
|
if isinstance(parameters, torch.Tensor):
|
|
|
|
parameters = [parameters]
|
|
|
|
for p in parameters:
|
|
|
|
if p.grad is None:
|
|
|
|
continue
|
|
|
|
if p.grad.device.type == 'cuda':
|
|
|
|
cuda_grads.append(p.grad.detach())
|
|
|
|
else:
|
|
|
|
cpu_grads.append(p.grad.detach())
|
|
|
|
if len(cuda_grads) > 0:
|
|
|
|
dummy_overflow_buf = torch.cuda.IntTensor([0])
|
2022-12-23 09:07:03 +00:00
|
|
|
multi_tensor_applier(fused_optim.multi_tensor_scale, dummy_overflow_buf, [cuda_grads, cuda_grads],
|
|
|
|
clip_coef)
|
2022-08-11 14:58:58 +00:00
|
|
|
for g in cpu_grads:
|
|
|
|
g.mul_(clip_coef)
|
|
|
|
|
|
|
|
|
|
|
|
def clip_grad_norm(parameters, max_norm: float, norm_type: float = 2.0) -> float:
|
|
|
|
total_norm = compute_grad_norm(parameters, norm_type)
|
|
|
|
_clip_grad_norm(parameters, max_norm, total_norm)
|
|
|
|
return total_norm
|
|
|
|
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
def clip_grad_norm_fp32(parameters, max_norm, norm_type=2):
|
2022-01-21 02:44:30 +00:00
|
|
|
"""Clips gradient norm of an iterable of parameters whose gradients are in fp32.
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
|
2022-01-21 02:44:30 +00:00
|
|
|
This is adapted from :func:`torch.nn.utils.clip_grad.clip_grad_norm_` and
|
2022-03-25 05:02:39 +00:00
|
|
|
added functionality to handle model parallel parameters.
|
|
|
|
|
|
|
|
Note:
|
|
|
|
the gradients are modified in place.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
parameters (Iterable[:class:`torch.tensor`] or :class:`torch.tensor`):
|
|
|
|
An iterable of Tensors or a single Tensor that will have gradients normalized.
|
|
|
|
max_norm (Union[float, int]): Max norm of the gradients.
|
|
|
|
norm_type (Union[float, int, 'inf']): Type of the used p-norm. Can be ``'inf'`` for infinity norm.
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
float: Total norm of the parameters.
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
"""
|
|
|
|
|
|
|
|
if isinstance(parameters, torch.Tensor):
|
|
|
|
parameters = [parameters]
|
|
|
|
|
|
|
|
# Filter parameters based on:
|
|
|
|
# - grad should not be none
|
|
|
|
# - parameter should not be shared
|
|
|
|
# - should not be a replica due to tensor model parallelism
|
2022-03-01 10:17:01 +00:00
|
|
|
params: List[Parameter] = []
|
|
|
|
has_zero_shared_param: bool = False
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
for param in parameters:
|
|
|
|
if param.grad is not None:
|
|
|
|
# Make sure the grads are in fp32
|
2022-03-01 10:17:01 +00:00
|
|
|
assert param.grad.dtype == torch.float, \
|
|
|
|
f'expected gradient to be dtype torch.float, but got {param.grad.type()}'
|
2022-07-08 05:34:48 +00:00
|
|
|
if hasattr(param, 'colo_attr') and param.colo_attr.sharded_data_tensor.is_sharded:
|
2022-03-01 10:17:01 +00:00
|
|
|
has_zero_shared_param = True
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
params.append(param)
|
2022-03-01 10:17:01 +00:00
|
|
|
|
|
|
|
if len(params) == 0:
|
2022-06-27 01:53:57 +00:00
|
|
|
enable_cuda_kernels = False
|
|
|
|
else:
|
|
|
|
enable_cuda_kernels = params[0].grad.device.type == 'cuda'
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
# Norm parameters.
|
|
|
|
max_norm = float(max_norm)
|
|
|
|
norm_type = float(norm_type)
|
|
|
|
|
2022-03-01 10:17:01 +00:00
|
|
|
# Parameters can be on CPU or CUDA
|
|
|
|
# If parameters are on CPU, disable CUDA kernerls
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
# Calculate norm.
|
|
|
|
if norm_type == inf:
|
|
|
|
total_norm = max(p.grad.data.abs().max() for p in params)
|
|
|
|
total_norm_cuda = torch.cuda.FloatTensor([float(total_norm)])
|
|
|
|
# Take max across all model-parallel GPUs.
|
2021-12-30 07:56:46 +00:00
|
|
|
if gpc.is_initialized(ParallelMode.MODEL) and gpc.get_world_size(ParallelMode.MODEL) > 1:
|
|
|
|
dist.all_reduce(total_norm_cuda,
|
|
|
|
op=dist.ReduceOp.MAX,
|
|
|
|
group=gpc.get_group(ParallelMode.MODEL),
|
|
|
|
async_op=False)
|
2022-03-01 10:17:01 +00:00
|
|
|
if has_zero_shared_param:
|
|
|
|
dist.all_reduce(total_norm_cuda,
|
|
|
|
op=dist.ReduceOp.MAX,
|
|
|
|
group=gpc.get_group(ParallelMode.DATA),
|
|
|
|
async_op=False)
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
total_norm = total_norm_cuda[0].item()
|
|
|
|
else:
|
|
|
|
tensor_parallel_grads = []
|
|
|
|
no_tensor_parallel_grads = []
|
2022-03-01 10:17:01 +00:00
|
|
|
zero_sharded_grads = []
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
for p in params:
|
|
|
|
if is_model_parallel_parameter(p):
|
2022-02-14 03:15:02 +00:00
|
|
|
reductor = (gpc.get_world_size(ParallelMode.TENSOR) / getattr(p, NUM_PARTITIONS))**(1 / norm_type)
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
tensor_parallel_grads.append(p.grad.data / reductor)
|
2022-07-08 05:34:48 +00:00
|
|
|
elif hasattr(p, 'colo_attr') and p.colo_attr.sharded_data_tensor.is_sharded:
|
2022-03-01 10:17:01 +00:00
|
|
|
zero_sharded_grads.append(p.grad.data)
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
else:
|
|
|
|
no_tensor_parallel_grads.append(p.grad.data)
|
2022-01-07 07:08:36 +00:00
|
|
|
|
2022-03-01 10:17:01 +00:00
|
|
|
if norm_type == 2.0 and enable_cuda_kernels:
|
2022-03-21 05:35:04 +00:00
|
|
|
tensor_parallel_norm = _calc_l2_norm(tensor_parallel_grads)**norm_type
|
|
|
|
no_tensor_parallel_norm = _calc_l2_norm(no_tensor_parallel_grads)**norm_type
|
|
|
|
zero_sharded_norm = _calc_l2_norm(zero_sharded_grads)**norm_type
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
else:
|
|
|
|
tensor_parallel_norm = _calc_lp(tensor_parallel_grads, norm_type)
|
2022-02-14 03:15:02 +00:00
|
|
|
no_tensor_parallel_norm = _calc_lp(no_tensor_parallel_grads, norm_type)
|
2022-03-01 10:17:01 +00:00
|
|
|
zero_sharded_norm = _calc_lp(zero_sharded_grads, norm_type)
|
2022-06-27 01:53:57 +00:00
|
|
|
# If norm is type of float, then we convert them into torch.Tensor.
|
|
|
|
tensor_parallel_norm = _get_tensor_norm(tensor_parallel_norm, enable_cuda_kernels)
|
|
|
|
no_tensor_parallel_norm = _get_tensor_norm(no_tensor_parallel_norm, enable_cuda_kernels)
|
|
|
|
zero_sharded_norm = _get_tensor_norm(zero_sharded_norm, enable_cuda_kernels)
|
2022-03-01 10:17:01 +00:00
|
|
|
# If grads are on CPU, the norms is also on CPU. Cast them to CUDA tensors
|
|
|
|
if not enable_cuda_kernels:
|
|
|
|
tensor_parallel_norm = _move_norm_to_cuda(tensor_parallel_norm)
|
|
|
|
no_tensor_parallel_norm = _move_norm_to_cuda(no_tensor_parallel_norm)
|
|
|
|
zero_sharded_norm = _move_norm_to_cuda(zero_sharded_norm)
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
# Sum across all model-parallel GPUs.
|
|
|
|
if gpc.is_initialized(ParallelMode.TENSOR) and len(tensor_parallel_grads) > 0:
|
2022-02-14 03:15:02 +00:00
|
|
|
dist.all_reduce(tensor_parallel_norm, op=dist.ReduceOp.SUM, group=gpc.get_group(ParallelMode.TENSOR))
|
2022-03-01 10:17:01 +00:00
|
|
|
# Sum across all zero sharded GPUs
|
|
|
|
if len(zero_sharded_grads) > 0:
|
|
|
|
dist.all_reduce(zero_sharded_norm, group=gpc.get_group(ParallelMode.DATA))
|
|
|
|
no_tensor_parallel_norm += zero_sharded_norm
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
total_norm = tensor_parallel_norm + no_tensor_parallel_norm
|
|
|
|
if gpc.is_initialized(ParallelMode.PIPELINE) and gpc.get_world_size(ParallelMode.PIPELINE) > 1:
|
2022-03-21 05:35:04 +00:00
|
|
|
dist.all_reduce(total_norm, op=dist.ReduceOp.SUM, group=gpc.get_group(ParallelMode.PIPELINE))
|
|
|
|
total_norm = total_norm**(1.0 / norm_type)
|
2022-03-01 10:17:01 +00:00
|
|
|
if torch.is_tensor(total_norm):
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
total_norm = total_norm.item()
|
|
|
|
|
|
|
|
# Scale.
|
|
|
|
clip_coeff = max_norm / (total_norm + 1.0e-6)
|
|
|
|
if clip_coeff < 1.0:
|
2022-03-01 10:17:01 +00:00
|
|
|
if enable_cuda_kernels:
|
|
|
|
grads = [p.grad.detach() for p in params]
|
|
|
|
dummy_overflow_buf = torch.cuda.IntTensor([0])
|
2022-12-23 09:07:03 +00:00
|
|
|
multi_tensor_applier(fused_optim.multi_tensor_scale, dummy_overflow_buf, [grads, grads], clip_coeff)
|
2022-03-01 10:17:01 +00:00
|
|
|
else:
|
|
|
|
for p in params:
|
|
|
|
p.grad.detach().mul_(clip_coeff)
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
return total_norm
|
|
|
|
|
|
|
|
|
|
|
|
def count_zeros_fp32(parameters):
|
|
|
|
if isinstance(parameters, torch.Tensor):
|
|
|
|
parameters = [parameters]
|
|
|
|
|
|
|
|
# Filter parameters based on:
|
|
|
|
# - grad should not be none
|
|
|
|
# - parameter should not be shared
|
|
|
|
# - should not be a replica due to tensor model parallelism
|
|
|
|
total_num_zeros = 0.0
|
|
|
|
for param in parameters:
|
|
|
|
grad_not_none = param.grad is not None
|
|
|
|
is_not_tp_duplicate = param_is_not_tensor_parallel_duplicate(param)
|
|
|
|
if grad_not_none and is_not_tp_duplicate:
|
|
|
|
grad = param.grad.detach()
|
|
|
|
num_zeros = grad.numel() - torch.count_nonzero(grad)
|
|
|
|
total_num_zeros = num_zeros + total_num_zeros
|
|
|
|
|
2022-01-20 05:44:51 +00:00
|
|
|
total_num_zeros = torch.IntTensor([int(total_num_zeros)]).cuda()
|
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
# Sum across all model-parallel GPUs.
|
|
|
|
ops = []
|
2022-02-14 03:15:02 +00:00
|
|
|
ops.append(
|
|
|
|
dist.all_reduce(total_num_zeros, op=dist.ReduceOp.SUM, group=gpc.get_group(ParallelMode.TENSOR), async_op=True))
|
2022-01-20 05:44:51 +00:00
|
|
|
if gpc.is_initialized(ParallelMode.PIPELINE):
|
2022-02-14 03:15:02 +00:00
|
|
|
ops.append(
|
|
|
|
dist.all_reduce(total_num_zeros,
|
|
|
|
op=dist.ReduceOp.SUM,
|
|
|
|
group=gpc.get_group(ParallelMode.PIPELINE),
|
|
|
|
async_op=True))
|
2022-01-20 05:44:51 +00:00
|
|
|
|
Develop/experiments (#59)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
* Split conv2d, class token, positional embedding in 2d, Fix random number in ddp
Fix convergence in cifar10, Imagenet1000
* Integrate 1d tensor parallel in Colossal-AI (#39)
* fixed 1D and 2D convergence (#38)
* optimized 2D operations
* fixed 1D ViT convergence problem
* Feature/ddp (#49)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* support torch ddp
* fix loss accumulation
* add log for ddp
* change seed
* modify timing hook
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* Feature/pipeline (#40)
* remove redundancy func in setup (#19) (#20)
* use env to control the language of doc (#24) (#25)
* Support TP-compatible Torch AMP and Update trainer API (#27)
* Add gradient accumulation, fix lr scheduler
* fix FP16 optimizer and adapted torch amp with tensor parallel (#18)
* fixed bugs in compatibility between torch amp and tensor parallel and performed some minor fixes
* fixed trainer
* Revert "fixed trainer"
This reverts commit 2e0b0b76990e8d4e337add483d878c0f61cf5097.
* improved consistency between trainer, engine and schedule (#23)
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
* add an example of ViT-B/16 and remove w_norm clipping in LAMB (#29)
* add explanation for ViT example (#35) (#36)
* optimize communication of pipeline parallel
* fix grad clip for pipeline
Co-authored-by: Frank Lee <somerlee.9@gmail.com>
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
* optimized 3d layer to fix slow computation ; tested imagenet performance with 3d; reworked lr_scheduler config definition; fixed launch args; fixed some printing issues; simplified apis of 3d layers (#51)
* Update 2.5d layer code to get a similar accuracy on imagenet-1k dataset
* update api for better usability (#58)
update api for better usability
Co-authored-by: 1SAA <c2h214748@gmail.com>
Co-authored-by: ver217 <lhx0217@gmail.com>
Co-authored-by: puck_WCR <46049915+WANG-CR@users.noreply.github.com>
Co-authored-by: binmakeswell <binmakeswell@gmail.com>
Co-authored-by: アマデウス <kurisusnowdeng@users.noreply.github.com>
Co-authored-by: BoxiangW <45734921+BoxiangW@users.noreply.github.com>
2021-12-09 07:08:29 +00:00
|
|
|
for req in ops:
|
|
|
|
req.wait()
|
|
|
|
total_num_zeros = total_num_zeros.item()
|
|
|
|
|
|
|
|
return total_num_zeros
|
|
|
|
|
|
|
|
|
|
|
|
def copy_tensor_parallel_attributes(src_tensor, dst_tensor):
|
|
|
|
for attr in TENSOR_PARALLEL_ATTRIBUTES:
|
|
|
|
if hasattr(src_tensor, attr):
|
|
|
|
val = getattr(src_tensor, attr)
|
|
|
|
setattr(dst_tensor, attr, val)
|
|
|
|
|
|
|
|
|
|
|
|
def param_is_not_tensor_parallel_duplicate(param):
|
2022-02-14 03:15:02 +00:00
|
|
|
return (hasattr(param, IS_TENSOR_PARALLEL) and getattr(param, IS_TENSOR_PARALLEL)) or (gpc.get_local_rank(
|
|
|
|
ParallelMode.TENSOR) == 0)
|
2021-12-20 15:26:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
@contextmanager
|
|
|
|
def switch_virtual_pipeline_parallel_rank(rank):
|
|
|
|
prev_rank = gpc.virtual_pipeline_parallel_rank
|
|
|
|
try:
|
|
|
|
gpc.set_virtual_pipeline_parallel_rank(rank)
|
|
|
|
yield
|
|
|
|
finally:
|
|
|
|
gpc.set_virtual_pipeline_parallel_rank(prev_rank)
|
2022-04-11 02:46:08 +00:00
|
|
|
|
|
|
|
|
|
|
|
def disposable(func: Callable) -> Callable:
|
|
|
|
executed = False
|
|
|
|
|
|
|
|
@functools.wraps(func)
|
|
|
|
def wrapper(*args, **kwargs):
|
|
|
|
nonlocal executed
|
|
|
|
if not executed:
|
|
|
|
executed = True
|
|
|
|
return func(*args, **kwargs)
|
|
|
|
|
|
|
|
return wrapper
|