Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / NVIDIA/framework-reproducibility issues and pull requests
#43 - Update tensorflow.md
Pull Request -
State: open - Opened by faisal3325 7 months ago
#42 - make table a bit easier to read
Pull Request -
State: closed - Opened by zivsha almost 2 years ago
#41 - `torch_status.md` determinism tracker similiar to `tensorflow_status.md`
Issue -
State: open - Opened by h-grieve over 2 years ago
- 1 comment
#40 - Changing class name/structure changes program functionality (using TensorFlow)
Issue -
State: closed - Opened by edwardyehuang over 2 years ago
- 2 comments
#39 - tf.data.experimental.sample_from_datasets non-deterministic in multi-gpu.
Issue -
State: open - Opened by yanniskar almost 3 years ago
- 4 comments
#38 - Non-reproducible model training results with TensorFlow 2.5
Issue -
State: closed - Opened by Brandhsu almost 3 years ago
- 7 comments
#37 - Early versions of TensorFlow (e.g. 1.9) do not have a version attribute
Issue -
State: open - Opened by SageAgastya over 3 years ago
- 3 comments
Labels: bug
#36 - Usage of numpy.random.Generator to deal with Data-Loader Parallelism
Issue -
State: closed - Opened by zakajd over 3 years ago
- 7 comments
#35 - garder14/byol-tensorflow2 (batch-norm & softmax/cross-entropy)
Issue -
State: open - Opened by evolu8 over 3 years ago
- 1 comment
#33 - Softmax xent
Pull Request -
State: open - Opened by wenscarl almost 4 years ago
#32 - patch tf.image.resize with bilinear by casting image to tf.float64
Issue -
State: closed - Opened by llan-ml almost 4 years ago
- 2 comments
#31 - More segment ops need to be patched
Issue -
State: closed - Opened by llan-ml about 4 years ago
- 5 comments
#30 - Model not deterministic/reproducible with all seeds set and os.environ['TF_DETERMINISTIC_OPS'] = '1'
Issue -
State: open - Opened by stevenwong951028 about 4 years ago
#29 - Don't get deterministic results with NGC 20.09
Issue -
State: open - Opened by ornellamarciano about 4 years ago
- 8 comments
#28 - Determinism across GPU architectures?
Issue -
State: closed - Opened by isarandi about 4 years ago
- 5 comments
Labels: question
#27 - Exception thrown with "No algorithm worked!" message on NGC 20.09
Issue -
State: closed - Opened by disembarrasing about 4 years ago
- 13 comments
#26 - Model nondeterminism when using `K.layers.DepthwiseConv2D` / `tf.nn.depthwise_conv2d` (known issue)
Issue -
State: open - Opened by kimzt about 4 years ago
- 18 comments
Labels: known issue
#25 - Message passing neural network determinism thwarted by tf.math.segment_sum and tf.gather
Issue -
State: closed - Opened by dangthatsright over 4 years ago
- 13 comments
#24 - non-determinism on tf.keras.layers.UpSampling2D(..., interpolation='nearest')
Issue -
State: closed - Opened by emiliocoutinho over 4 years ago
- 3 comments
#23 - Fix typo in main/TensorFlow README
Pull Request -
State: closed - Opened by SeonbeomKim over 4 years ago
- 1 comment
#22 - Running on stock TensorFlow version >= 2.1
Issue -
State: closed - Opened by emiliocoutinho over 4 years ago
- 5 comments
#21 - [WIP] Feature/patch/softmax cross entropy with logits
Pull Request -
State: open - Opened by MFreidank over 4 years ago
- 14 comments
#20 - Make tf_determinism.patch() handle tf>=2.1 as well
Issue -
State: closed - Opened by MFreidank over 4 years ago
- 4 comments
#19 - Reproducibility issue with transformers (BERT) and tf2.2
Issue -
State: open - Opened by MFreidank over 4 years ago
- 21 comments
#18 - Nondeterminism from tf.image.crop_and_resize
Issue -
State: closed - Opened by xthyax over 4 years ago
- 2 comments
#17 - Added non-determinism by unsorted_segment_sum
Pull Request -
State: closed - Opened by gavins13 over 4 years ago
- 1 comment
#16 - Inc. solution for AutoGraph loop conversion
Pull Request -
State: closed - Opened by gavins13 over 4 years ago
- 6 comments
#15 - Add layers.Dropout as another random seed to set
Pull Request -
State: closed - Opened by ned2 over 4 years ago
- 2 comments
#14 - Lack of reproducibility when using Huggingface transformers library (TensorFlow version)
Issue -
State: open - Opened by dmitriydligach over 4 years ago
- 11 comments
Labels: debugging
#13 - Will this package work on tf.keras model?
Issue -
State: closed - Opened by leocnj over 4 years ago
- 1 comment
Labels: question
#12 - How can I get deterministic, back-propagatable upsampling?
Issue -
State: closed - Opened by louislbc almost 5 years ago
- 5 comments
Labels: question
#11 - Update method call to set TF random seed
Pull Request -
State: closed - Opened by Simsso almost 5 years ago
- 1 comment
#10 - Are there disadvantages to enabling deterministic ops?
Issue -
State: closed - Opened by lminer almost 5 years ago
- 2 comments
Labels: question
#9 - Getting OpenNMT-tf to train reproducibly
Issue -
State: open - Opened by atebbifakhr almost 5 years ago
- 13 comments
Labels: debugging, possible new discovery
#8 - Support for TF 2.1.0
Issue -
State: closed - Opened by jmwoloso almost 5 years ago
- 3 comments
Labels: enhancement
#7 - Raise more specific exceptions
Issue -
State: closed - Opened by bersbersbers almost 5 years ago
- 3 comments
Labels: enhancement
#6 - How can I check whether I'm in a NGC TensorFlow container?
Issue -
State: closed - Opened by AndreaPi almost 5 years ago
- 2 comments
Labels: question
#5 - Tensorflow-determinism not working with conda tensorflow-gpu
Issue -
State: closed - Opened by Eugen2525 about 5 years ago
- 6 comments
Labels: wontfix
#4 - Why tensorflow-determinism and not simply determinism?
Issue -
State: closed - Opened by kirk86 about 5 years ago
- 4 comments
Labels: question
#3 - Unidentified model not training reproducibly
Issue -
State: closed - Opened by invoxiaehu about 5 years ago
- 8 comments
Labels: waiting for code
#2 - How can I confirm that the deterministic environment variable is working?
Issue -
State: closed - Opened by cklsoft about 5 years ago
- 5 comments
Labels: question
#1 - When will the debug tool be available?
Issue -
State: closed - Opened by salinaaaaaa over 5 years ago
- 2 comments
Labels: question