-
Notifications
You must be signed in to change notification settings - Fork 326
Issues: tensorflow/model-optimization
Sparsity Runtime Integration with TF/TFLite for Latency Impro...
#173
opened Dec 6, 2019 by
alanchiao
Open
30
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Activity Regularizer not working with quantization aware training (QAT) TF 2.6
#842
opened Sep 27, 2021 by
bayesian-mind
Make Model Optimization Project Contributor Friendly
feature request
feature request
technique:all
Applicable to all techniques and TFMOT collectively.
#131
opened Oct 21, 2019 by
alanchiao
5 of 6 tasks
Unable to get tfmot version from API
bug
Something isn't working
#594
opened Nov 17, 2020 by
Lotte1990
quantize_model and tf.keras.mixed_precision.experimental.Policy("mixed_float16") throws ValueError
bug
Something isn't working
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#367
opened Apr 18, 2020 by
HauserA
raise FailedPreconditionError while training model
bug
Something isn't working
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#46
opened Jul 29, 2019 by
johnwang416
Object detection API
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#132
opened Oct 22, 2019 by
dbrazey
Opensource Mobilenet V1 Image Classification Training Script
feature request
feature request
#133
opened Oct 23, 2019 by
alanchiao
Pruning: Keras subclassed model increased support
contributions welcome
See CONTRIBUTING.md for details.
feature request
feature request
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#155
opened Nov 11, 2019 by
alanchiao
Sparsity Runtime Integration with TF/TFLite for Latency Improvements
feature request
feature request
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#173
opened Dec 6, 2019 by
alanchiao
Pruning: Checkpointed Models Not Guaranteed To Be Sparse
contributions welcome
See CONTRIBUTING.md for details.
feature request
feature request
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#206
opened Dec 30, 2019 by
alanchiao
Pruning fails for siamese networks
feature request
feature request
priority:low
Low priority when applied. Intentionally open with no assignee or contributors welcome label.
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#270
opened Feb 26, 2020 by
digitalheir
Pruning: improve custom training loop API
feature request
feature request
priority:low
Low priority when applied. Intentionally open with no assignee or contributors welcome label.
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#271
opened Feb 27, 2020 by
alanchiao
Move TFMOT Library Code Off Private APIs
technique:all
Applicable to all techniques and TFMOT collectively.
#305
opened Mar 18, 2020 by
alanchiao
Support for Conv1D Quantization
feature request
feature request
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#362
opened Apr 15, 2020 by
madarax64
How to disable change the output tensor name?
feature request
feature request
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#28
opened Jun 30, 2019 by
zhen8838
Tracking TFMOT 1.0.0 Release
technique:all
Applicable to all techniques and TFMOT collectively.
#371
opened Apr 22, 2020 by
alanchiao
QAT (quantization aware training) Support quantizing models recursively
feature request
feature request
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#377
opened May 4, 2020 by
CRosero
H5 to Pb Conversion with Fake Quantization Node Fails
bug
Something isn't working
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#381
opened May 11, 2020 by
anidh
Behaviour of stripped models and sequence masking
bug
Something isn't working
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#403
opened May 21, 2020 by
captainproton1971
sparsity.prune_low_magnitude fails with mixed precision policy mixed_float16
feature request
feature request
technique:pruning
Regarding tfmot.sparsity.keras APIs and docs
#409
opened May 28, 2020 by
dsuthar-nvidia
[Feature request or potential bug] Override of default default_8bit_quantize_layout_transform
feature request
feature request
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#445
opened Jun 30, 2020 by
alessandroaimar
How to user quantize to imporve inference performance on tensorflow-serving?
bug
Something isn't working
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#450
opened Jul 2, 2020 by
ZhiyiLan
Post quantization does not utilize GPU
feature request
feature request
#454
opened Jul 8, 2020 by
ek9852
Support Dense+BatchNorm in QAT
bug
Something isn't working
technique:qat
Regarding tfmot.quantization.keras (for quantization-aware training) APIs and docs
#363
opened Apr 16, 2020 by
lravano
Previous Next
ProTip!
Follow long discussions with comments:>50.