cybertronai/gradient-checkpointing issues and pull requests

#57 - How does gradient checkpointing relate to reversible layers?

Issue - State: closed - Opened by styleex 9 months ago - 2 comments

#56 - Publish this on Pypi

Issue - State: closed - Opened by LucaCappelletti94 over 3 years ago - 2 comments

#55 - Speed mode is much slower than memory mode

Issue - State: open - Opened by jasperzhong over 3 years ago

#54 - tf.GradientTape and higher order derivatives

Issue - State: open - Opened by ilyak93 over 3 years ago

#53 - Limiting memory usage via GPUOptions conflicts with is_gpu_available

Issue - State: closed - Opened by lmatz almost 4 years ago

#52 - TF while loop error

Issue - State: open - Opened by akikaaa almost 4 years ago - 1 comment

#51 - is this module adapted for tensorflow 2.0 and above?

Issue - State: open - Opened by sohamhans19994 over 4 years ago - 1 comment

#50 - Solving the conflict with Keras BatchNormalization.

Pull Request - State: open - Opened by purpleyun over 4 years ago

#49 - Does this package work for tensorflow 1.15?

Issue - State: open - Opened by Jingnan-Jia almost 5 years ago - 7 comments

#48 - Package Maintenance for TF-2.0 with Contrib Module Sunset

Issue - State: closed - Opened by Uiuran about 5 years ago - 10 comments

#47 - Gradient checkpointing seems to conflict with Keras batch norm

Issue - State: open - Opened by demmerichs about 5 years ago - 9 comments

#46 - Some high level descriptions on the MEMORY policy will be very helpful

Issue - State: open - Opened by wangli1426 over 5 years ago - 4 comments

#45 - Update suggested usage

Pull Request - State: closed - Opened by yselivonchyk over 5 years ago - 1 comment

#44 - `mem_util` only shows CPU, not GPU

Issue - State: open - Opened by ghost over 5 years ago

#43 - Fix OOM when splitting model across multiple GPUs

Pull Request - State: closed - Opened by yselivonchyk over 5 years ago - 1 comment

#42 - Does Not Work with Keras

Issue - State: open - Opened by ghost over 5 years ago - 66 comments

#41 - Does this work with `fit_generator`?

Issue - State: open - Opened by ghost over 5 years ago - 4 comments

#40 - Grad CAM Map With Memory Saving Looks Odd

Issue - State: open - Opened by ghost over 5 years ago - 3 comments

#39 - Rerunning from Checkpoint Gives Error

Issue - State: open - Opened by ghost over 5 years ago - 2 comments

#38 - Splitting model across 2 GPUs leads to OOM

Issue - State: closed - Opened by yselivonchyk over 5 years ago - 9 comments

#37 - Added setup example

Pull Request - State: closed - Opened by yselivonchyk over 5 years ago

#36 - update README with repo status

Pull Request - State: closed - Opened by christopherhesse almost 6 years ago

#35 - Does this lib work with layers that contain randomness?

Issue - State: open - Opened by Lyken17 almost 6 years ago

#34 - copy device info when copying ops

Pull Request - State: closed - Opened by ChenChuang about 6 years ago - 1 comment

#33 - Extremely slow when running on distributed tensorflow with horovod

Issue - State: closed - Opened by ChenChuang about 6 years ago - 2 comments

#32 - Problems with custom gradient

Issue - State: open - Opened by artem-s-shevchenko about 6 years ago - 1 comment

#31 - add timeline module

Pull Request - State: closed - Opened by justheuristic about 6 years ago

#30 - handle sparse gradients

Pull Request - State: closed - Opened by justheuristic about 6 years ago - 1 comment

#29 - Doesn't seemingly work with latest TF versions.

Issue - State: open - Opened by joeyearsley about 6 years ago - 19 comments

#28 - Checkpointing with collections raising exception in Keras

Issue - State: open - Opened by skottapa over 6 years ago - 1 comment

#27 - 'NoneType' object has no attribute 'op'

Issue - State: open - Opened by siavash-khodadadeh over 6 years ago - 1 comment

#26 - How to select nodes for collections

Issue - State: closed - Opened by jianlong-yuan over 6 years ago - 1 comment

#25 - Update requirements.txt

Pull Request - State: closed - Opened by diego- over 6 years ago

#24 - Cleared up TF warning.

Pull Request - State: closed - Opened by thomasquintana over 6 years ago - 1 comment

#23 - Update memory_saving_gradients.py

Pull Request - State: closed - Opened by dmitrivainbrand over 6 years ago - 1 comment

#22 - Checkpointing of VGG

Issue - State: open - Opened by dmitrivainbrand over 6 years ago

#21 - Checkpointing for FP16

Issue - State: open - Opened by dmitrivainbrand over 6 years ago - 1 comment

#20 - Update README to include more related work

Pull Request - State: closed - Opened by davidBelanger over 6 years ago - 1 comment

#19 - A few more citations

Issue - State: open - Opened by davidBelanger over 6 years ago - 1 comment

#19 - A few more citations

Issue - State: open - Opened by davidBelanger over 6 years ago - 1 comment

#18 - OOM when using gradients_memory's list of checkpointed tensors

Issue - State: open - Opened by IdoZach over 6 years ago - 1 comment

#17 - TF 1.6rc1 Support

Issue - State: closed - Opened by michaelklachko over 6 years ago - 3 comments

#17 - TF 1.6rc1 Support

Issue - State: closed - Opened by michaelklachko over 6 years ago - 3 comments

#16 - memory_test.py failed

Issue - State: closed - Opened by majia-yu almost 7 years ago - 2 comments

#16 - memory_test.py failed

Issue - State: closed - Opened by majia-yu almost 7 years ago - 2 comments

#15 - fix fwd_ops debug_print that was printing bwd_ops instead of fwd_ops

Pull Request - State: closed - Opened by ethancaballero almost 7 years ago

#15 - fix fwd_ops debug_print that was printing bwd_ops instead of fwd_ops

Pull Request - State: closed - Opened by ethancaballero almost 7 years ago

#14 - how to find out which checkpoint nodes were selected with memory heuristic

Issue - State: open - Opened by lexbee almost 7 years ago - 1 comment

#14 - how to find out which checkpoint nodes were selected with memory heuristic

Issue - State: open - Opened by lexbee almost 7 years ago - 1 comment

#13 - Use with static (unrolled) RNN?

Issue - State: open - Opened by eundersander almost 7 years ago - 4 comments

#12 - core dump

Issue - State: open - Opened by shiyongde almost 7 years ago - 1 comment

#12 - core dump

Issue - State: open - Opened by shiyongde almost 7 years ago - 1 comment

#11 - gradients_memory require more memory than tf.Optimizer.minimize

Issue - State: open - Opened by gchlebus almost 7 years ago - 12 comments

#10 - Wrong timeline computation

Issue - State: closed - Opened by gchlebus almost 7 years ago - 4 comments

#10 - Wrong timeline computation

Issue - State: closed - Opened by gchlebus almost 7 years ago - 4 comments

#9 - AttributeError: 'NoneType' object has no attribute 'pred'

Issue - State: open - Opened by Mohamedemad4 almost 7 years ago - 3 comments

#9 - AttributeError: 'NoneType' object has no attribute 'pred'

Issue - State: open - Opened by Mohamedemad4 almost 7 years ago - 3 comments

#8 - Using gradient checkpointing with a optimizer

Issue - State: open - Opened by ryx2 almost 7 years ago - 1 comment

#8 - Using gradient checkpointing with a optimizer

Issue - State: open - Opened by ryx2 almost 7 years ago - 1 comment

#7 - RNN while loop context error

Issue - State: closed - Opened by maxosmith almost 7 years ago - 4 comments

#7 - RNN while loop context error

Issue - State: closed - Opened by maxosmith almost 7 years ago - 4 comments

#6 - Is it possible to implement in dlib?

Issue - State: closed - Opened by chikiuso almost 7 years ago - 1 comment

#6 - Is it possible to implement in dlib?

Issue - State: closed - Opened by chikiuso almost 7 years ago - 1 comment

#5 - Cannot fit any extra batches into memory than normal

Issue - State: closed - Opened by EmilHJ almost 7 years ago - 8 comments

#4 - How do you use gradient checkpointing if you are updating gradients with an optimizer?

Issue - State: closed - Opened by ryx2 almost 7 years ago - 2 comments

#3 - Use flake8 to find syntax errors, undefined names

Pull Request - State: closed - Opened by cclauss almost 7 years ago

#3 - Use flake8 to find syntax errors, undefined names

Pull Request - State: closed - Opened by cclauss almost 7 years ago

#2 - Don’t forget self

Pull Request - State: closed - Opened by cclauss almost 7 years ago

#2 - Don’t forget self

Pull Request - State: closed - Opened by cclauss almost 7 years ago

#1 - Optimal checkpointing of reverse differentiation has been already solved

Issue - State: closed - Opened by bdauvergne almost 7 years ago - 1 comment

GitHub / cybertronai/gradient-checkpointing issues and pull requests