You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ColossalAI/colossalai/legacy/trainer/hooks/_log_hook.py

303 lines
13 KiB

3 years ago
#!/usr/bin/env python
# -*- encoding: utf-8 -*-
import os
import os.path as osp
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>
3 years ago
from typing import List
from colossalai.legacy.context import ParallelMode
from colossalai.legacy.core import global_context as gpc
from colossalai.legacy.registry import HOOKS
from colossalai.legacy.trainer.hooks._metric_hook import ThroughputMetric
from colossalai.legacy.utils import is_dp_rank_0, is_no_pp_or_last_stage, is_tp_rank_0, report_memory_usage
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>
3 years ago
from colossalai.logging import DistributedLogger
from colossalai.utils import MultiTimer
from ._base_hook import BaseHook
from ._commons_ import _format_number
3 years ago
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>
3 years ago
class LogByEpochHook(BaseHook):
"""Hook to log by epoch.
Args:
logger (:class:`colossalai.logging.DistributedLogger`): Logger for recording the log information.
interval (int, optional): Interval of printing log information, defaults to 1.
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front,
defaults to 1. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
"""
def __init__(self, logger, interval: int = 1, priority: int = 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>
3 years ago
super().__init__(priority)
self.logger = logger
3 years ago
self._interval = interval
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>
3 years ago
def _is_epoch_to_log(self, trainer):
return trainer.cur_epoch % self._interval == 0
3 years ago
@HOOKS.register_module
class LogMetricByStepHook(BaseHook):
"""Hook to log metric by step.
Args:
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front,
defaults to 10. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
"""
def __init__(self, priority: int = 10):
super().__init__(priority)
def after_train_iter(self, trainer, *args):
trainer.states["step_metrics"] = dict()
for metric_name, metric_calculator in trainer.states["metrics"]["train"].items():
if isinstance(metric_calculator, ThroughputMetric):
trainer.states["step_metrics"][metric_name.lower()] = metric_calculator.get_last_step_info()
else:
trainer.states["step_metrics"][metric_name.lower()] = metric_calculator.get_last_step_value()
def after_test_iter(self, trainer, *args):
trainer.states["step_metrics"] = dict()
for metric_name, metric_calculator in trainer.states["metrics"]["test"].items():
if isinstance(metric_calculator, ThroughputMetric):
trainer.states["step_metrics"][metric_name.lower()] = metric_calculator.get_last_step_info()
else:
trainer.states["step_metrics"][metric_name.lower()] = metric_calculator.get_last_step_value()
3 years ago
@HOOKS.register_module
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>
3 years ago
class LogMetricByEpochHook(LogByEpochHook):
"""Specialized hook to record the metric to log.
3 years ago
Args:
logger (:class:`colossalai.logging.DistributedLogger`): Logger for recording the log information.
interval (int, optional): Interval of printing log information, defaults to 1.
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front,
defaults to 10. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
3 years ago
"""
def __init__(self, logger, interval: int = 1, priority: int = 10) -> None:
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>
3 years ago
super().__init__(logger, interval, priority)
3 years ago
self._is_rank_to_log = is_dp_rank_0() and is_tp_rank_0() and 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>
3 years ago
def _get_str(self, trainer, mode):
3 years ago
msg = []
for metric_name, metric_calculator in trainer.states["metrics"][mode].items():
msg.append(f"{metric_name} = {_format_number(metric_calculator.get_accumulated_value())}")
msg = " | ".join(msg)
3 years ago
return msg
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>
3 years ago
def after_train_epoch(self, trainer):
if self._is_epoch_to_log(trainer):
msg = self._get_str(trainer=trainer, mode="train")
3 years ago
if self._is_rank_to_log:
self.logger.info(f"[Epoch {trainer.cur_epoch} / Train]: {msg}")
# f'Training - Epoch {trainer.cur_epoch} - {self.__class__.__name__}: {msg}')
3 years ago
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>
3 years ago
def after_test_epoch(self, trainer):
if self._is_epoch_to_log(trainer):
msg = self._get_str(trainer=trainer, mode="test")
3 years ago
if self._is_rank_to_log:
self.logger.info(f"[Epoch {trainer.cur_epoch} / Test]: {msg}")
# f'Testing - Epoch {trainer.cur_epoch} - {self.__class__.__name__}: {msg}')
3 years ago
@HOOKS.register_module
class TensorboardHook(BaseHook):
"""Specialized hook to record the metric to Tensorboard.
3 years ago
Args:
log_dir (str): Directory of log.
ranks (list): Ranks of processors.
parallel_mode (:class:`colossalai.legacy.context.parallel_mode.ParallelMode`, optional): Parallel mode used in trainer,
defaults to colossalai.legacy.context.parallel_mode.ParallelMode.GLOBAL.
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front,
defaults to 10. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
3 years ago
"""
def __init__(
self,
log_dir: str,
ranks: List = None,
parallel_mode: ParallelMode = ParallelMode.GLOBAL,
priority: int = 10,
) -> None:
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>
3 years ago
super().__init__(priority=priority)
from torch.utils.tensorboard import SummaryWriter
3 years ago
# create log dir
if not gpc.is_initialized(ParallelMode.GLOBAL) or gpc.get_global_rank() == 0:
os.makedirs(log_dir, exist_ok=True)
# determine the ranks to generate tensorboard logs
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>
3 years ago
self._is_valid_rank_to_log = False
if not gpc.is_initialized(parallel_mode):
self._is_valid_rank_to_log = True
else:
local_rank = gpc.get_local_rank(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>
3 years ago
if ranks is None or local_rank in ranks:
self._is_valid_rank_to_log = 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>
3 years ago
# check for
if (
gpc.is_initialized(ParallelMode.PIPELINE)
and not gpc.is_last_rank(ParallelMode.PIPELINE)
and self._is_valid_rank_to_log
):
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>
3 years ago
raise ValueError("Tensorboard hook can only log on the last rank of pipeline process group")
if self._is_valid_rank_to_log:
3 years ago
# create workspace on only one rank
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>
3 years ago
if gpc.is_initialized(parallel_mode):
rank = gpc.get_local_rank(parallel_mode)
3 years ago
else:
rank = 0
# create workspace
log_dir = osp.join(log_dir, f"{parallel_mode}_rank_{rank}")
os.makedirs(log_dir, exist_ok=True)
3 years ago
self.writer = SummaryWriter(log_dir=log_dir, filename_suffix=f"_rank_{rank}")
3 years ago
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>
3 years ago
def _log_by_iter(self, trainer, mode: str):
for metric_name, metric_calculator in trainer.states["metrics"][mode].items():
3 years ago
if metric_calculator.epoch_only:
continue
val = metric_calculator.get_last_step_value()
if self._is_valid_rank_to_log:
self.writer.add_scalar(f"{metric_name}/{mode}", val, trainer.cur_step)
3 years ago
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>
3 years ago
def _log_by_epoch(self, trainer, mode: str):
for metric_name, metric_calculator in trainer.states["metrics"][mode].items():
3 years ago
if metric_calculator.epoch_only:
val = metric_calculator.get_accumulated_value()
if self._is_valid_rank_to_log:
self.writer.add_scalar(f"{metric_name}/{mode}", val, trainer.cur_step)
3 years ago
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>
3 years ago
def after_test_iter(self, trainer, *args):
self._log_by_iter(trainer, mode="test")
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>
3 years ago
def after_test_epoch(self, trainer):
self._log_by_epoch(trainer, mode="test")
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>
3 years ago
def after_train_iter(self, trainer, *args):
self._log_by_iter(trainer, mode="train")
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>
3 years ago
def after_train_epoch(self, trainer):
self._log_by_epoch(trainer, mode="train")
3 years ago
@HOOKS.register_module
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>
3 years ago
class LogTimingByEpochHook(LogByEpochHook):
"""Specialized hook to write timing record to log.
3 years ago
Args:
timer (:class:`colossalai.utils.MultiTimer`): Timer for the hook.
logger (:class:`colossalai.logging.DistributedLogger`): Logger for recording the log information.
interval (int, optional): Interval of printing log information, defaults to 1.
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front
defaults to 10. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
log_eval (bool, optional): Whether writes in evaluation, defaults to True.
ignore_num_train_steps (int, optional): Number of training steps to ignore, defaults to 0.
3 years ago
"""
def __init__(
self,
timer: MultiTimer,
logger: DistributedLogger,
interval: int = 1,
priority: int = 10,
log_eval: bool = True,
ignore_num_train_steps: int = 0,
) -> None:
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>
3 years ago
super().__init__(logger=logger, interval=interval, priority=priority)
self._timer = timer
3 years ago
self._log_eval = log_eval
self._is_rank_to_log = is_dp_rank_0() and is_tp_rank_0() and is_no_pp_or_last_stage()
3 years ago
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>
3 years ago
# extra handling to avoid the unstable readings of the first
# few training steps to affect the history mean time
self._ignore_num_train_steps = ignore_num_train_steps
self._is_train_step_history_trimmed = False
def _get_message(self, mode):
3 years ago
msg = []
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>
3 years ago
for timer_name, timer in self._timer:
if timer_name.startswith(mode):
last_elapsed_time = timer.get_elapsed_time()
if timer.has_history:
if timer_name == "Train-step" and not self._is_train_step_history_trimmed:
timer._history = timer._history[self._ignore_num_train_steps :]
self._is_train_step_history_trimmed = True
history_mean = timer.get_history_mean()
timer.get_history_sum()
msg.append(
f"{timer_name}: last = {_format_number(last_elapsed_time)} s, mean = {_format_number(history_mean)} s"
)
else:
msg.append(f"{timer_name}: last = {_format_number(last_elapsed_time)} s")
msg = " | ".join(msg)
3 years ago
return msg
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>
3 years ago
def after_train_epoch(self, trainer):
"""Writes log after finishing a training epoch."""
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>
3 years ago
if self._is_epoch_to_log(trainer) and self._is_rank_to_log:
msg = self._get_message("Train")
self.logger.info(f"[Epoch {trainer.cur_epoch} / Train]: {msg} | #steps/epoch = {trainer.steps_per_epoch}")
3 years ago
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>
3 years ago
def after_test_epoch(self, trainer):
"""Writes log after finishing a testing epoch."""
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>
3 years ago
if self._is_epoch_to_log(trainer) and self._is_rank_to_log and self._log_eval:
msg = self._get_message("Test")
self.logger.info(f"[Epoch {trainer.cur_epoch} / Test]: {msg}")
3 years ago
@HOOKS.register_module
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>
3 years ago
class LogMemoryByEpochHook(LogByEpochHook):
3 years ago
"""Specialized Hook to write memory usage record to log.
Args:
logger (:class:`colossalai.logging.DistributedLogger`): Logger for recording the log information.
interval (int, optional): Interval of printing log information, defaults to 1.
priority (int, optional): Priority in the printing, hooks with small priority will be printed in front
defaults to 1. If different hooks share same priority, the order of printing would
depend on the hooks order in the hook list.
log_eval (bool, optional): Whether writes in evaluation, defaults to True.
3 years ago
"""
def __init__(
self,
logger: DistributedLogger,
interval: int = 1,
priority: int = 10,
log_eval: bool = True,
report_cpu: bool = False, # no reference
) -> None:
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>
3 years ago
super().__init__(logger=logger, interval=interval, priority=priority)
3 years ago
self._log_eval = log_eval
self._is_rank_to_log = is_dp_rank_0() and is_tp_rank_0()
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>
3 years ago
def before_train(self, trainer):
"""Resets before training."""
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>
3 years ago
if self._is_epoch_to_log(trainer) and self._is_rank_to_log:
report_memory_usage("Before-train", self.logger)
3 years ago
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>
3 years ago
def after_train_epoch(self, trainer):
"""Writes log after finishing a training epoch."""
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>
3 years ago
if self._is_epoch_to_log(trainer) and self._is_rank_to_log:
report_memory_usage(f"[Epoch {trainer.cur_epoch} / Train]", self.logger)
3 years ago
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>
3 years ago
def after_test(self, trainer):
"""Reports after testing."""
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>
3 years ago
if self._is_epoch_to_log(trainer) and self._is_rank_to_log and self._log_eval:
report_memory_usage(f"[Epoch {trainer.cur_epoch} / Test]", self.logger)