Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / OpenNMT/OpenNMT-tf issues and pull requests

#1011 - Support for Tensorflow 2.14

Pull Request - State: open - Opened by jordimas 3 months ago

#1010 - Remove tensorflow.addons dependency and bring needed classes

Pull Request - State: open - Opened by jordimas 3 months ago

#1009 - Multi-source NMT

Issue - State: open - Opened by ThuraAung1601 7 months ago

#1008 - Feature Request: Support for TensorFlow 2.14 and Higher

Issue - State: open - Opened by ncicio 8 months ago - 1 comment

#1006 - OpenNMT-tf onnx export ?

Issue - State: open - Opened by flutter-painter 10 months ago

#1005 - tensorflow lite model example

Issue - State: closed - Opened by flutter-painter 11 months ago - 8 comments

#1004 - Input size mismatch

Issue - State: open - Opened by pranjaliseth 12 months ago

#1003 - Fixes typos in configuration.md

Pull Request - State: open - Opened by jordimas about 1 year ago

#1002 - Support TensorFlow 2.13

Pull Request - State: closed - Opened by guillaumekln over 1 year ago

#1001 - word-level knowledge distillation

Issue - State: open - Opened by ZenBel over 1 year ago

#1000 - xla_ops failed when use multi gpu

Issue - State: open - Opened by wangshauitj over 1 year ago

#999 - Support TensorFlow 2.12

Pull Request - State: closed - Opened by guillaumekln over 1 year ago

#998 - Quickstart example runs for a long time

Issue - State: open - Opened by cspink over 1 year ago - 5 comments

#996 - Make timeout value configurable while searching for an optimal batch size

Issue - State: closed - Opened by sandeepchittilla over 1 year ago
Labels: enhancement

#995 - Fix GitHub Actions warnings

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#994 - Relax the sacrebleu version requirement

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#993 - Report the total number of tokens per second in the log (src+tgt)

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#992 - Add tests for recent changes related to the sample size

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#991 - Add jit_compile flag to compile the model with XLA

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#990 - Use a GradientTape in both eager and graph execution

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#989 - Fix some tests that are failing with Numpy 1.24

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#988 - Divide the gradients by the actual sample size

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#987 - Support TensorFlow 2.11

Pull Request - State: closed - Opened by guillaumekln almost 2 years ago

#986 - Fixed issue with horovod and early stopping

Pull Request - State: closed - Opened by awery almost 2 years ago - 1 comment

#985 - "Horovod has been shut down" error when training is finished due to early stopping

Issue - State: closed - Opened by awery almost 2 years ago - 2 comments
Labels: bug

#984 - chrF and chrF++

Pull Request - State: closed - Opened by awery almost 2 years ago - 1 comment

#983 - Support CTranslate2 3.0

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#982 - Fix error when using guided alignment with mixed precision

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#981 - Training fails with --mixed_precision and guided alignments

Issue - State: closed - Opened by Dagamies about 2 years ago
Labels: bug

#980 - ModuleNotFoundError: No module named 'lxml.etree'

Issue - State: closed - Opened by atanumandal0491 about 2 years ago - 1 comment

#979 - An input error occurred while expanding the dataset

Issue - State: closed - Opened by copperdong about 2 years ago - 1 comment

#978 - Remove recording of the global gradient norm

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#977 - Add flag to pad each batch to the length bucket boundary

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#976 - Do not consider parameter decay_step_duration as a V1 only parameter

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#975 - Remove non needed FP32 cast before softmax

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#974 - Fix count_lines function when the file is gzipped

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#973 - Parallel text inputter fails with compressed text files

Issue - State: closed - Opened by awery about 2 years ago - 1 comment
Labels: bug

#972 - Add Transformer configurations used in the Scaling NMT paper

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#971 - Add parameter to automatically share embeddings when possible

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#970 - Modify loss function or mask attention in this code

Issue - State: closed - Opened by madhurapande19 about 2 years ago - 1 comment
Labels: question

#969 - Accept list of checkpoints in average_checkpoints method

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#968 - Support TensorFlow 2.10

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#967 - Use new API to resolve variable names in checkpoints

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#966 - Use minimal gradient accumulation in batch size autotuning

Pull Request - State: closed - Opened by guillaumekln about 2 years ago

#965 - Fine-tune GPT2 model from Transformers

Issue - State: closed - Opened by NeonBohdan about 2 years ago - 4 comments

#964 - What material did you use to train your pretrained model?

Issue - State: closed - Opened by junyizhao04 about 2 years ago - 1 comment

#963 - Fix deprecation warnings for using distutils.version.LooseVersion

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#961 - Make initial learning rate of InvSqrtDecay configurable

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#960 - Include sacrebleu==2.2.* in requirements

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#959 - Avoid building batches with more tokens than the user batch size

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#958 - Add log when the averaged checkpoint is saved

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#957 - SequenceClassifier doesn't seem to learn leaked target

Issue - State: closed - Opened by Ingvar-Y over 2 years ago - 2 comments

#956 - cpp for tensorflow serving

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

#955 - Add argument to disable the bias in the output linear layer

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#954 - Add argument to disable bias terms in the multi-head attention

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#953 - Fix dtype of SequenceRecordInputter's length feature

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#952 - An issue with SequenceRecordInputter ?

Issue - State: closed - Opened by Ingvar-Y over 2 years ago - 1 comment
Labels: bug

#951 - WIP : First implementation of sliding window local sparse attention.

Pull Request - State: open - Opened by natsegal over 2 years ago

#950 - Fix evaluation and scoring with language models

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#949 - Use simpler structure for the generated API reference

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#948 - Fix inference features returned by the language model inputter

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#947 - Set a minimum elapsed time of 1 ms for logging metrics

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#946 - Remove support for older TensorFlow versions 2.4 & 2.5

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#945 - Integrate the new CTranslate2 converter for OpenNMT-tf

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#944 - Support TensorFlow 2.9

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#943 - Error when inferencing with gpt.

Issue - State: closed - Opened by songxxzp over 2 years ago - 3 comments
Labels: bug

#942 - Rollback to tf.nn.embedding_lookup in non TensorFlow Lite mode

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#941 - Why BLEU score is computed during training and not during inference?

Issue - State: closed - Opened by code-rex1 over 2 years ago - 1 comment

#940 - Update docs requirements

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#939 - Running sequence tagger using OpenNMT-tf

Issue - State: closed - Opened by faysalhossain2007 over 2 years ago - 2 comments

#938 - Fix documentation build

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#937 - Can I change opennmt-py model to opennmt-tf model?

Issue - State: closed - Opened by 520jefferson over 2 years ago - 4 comments
Labels: question

#935 - Clarify the inference buffering log

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#934 - Fix WordNoiser on inputs with rank > 2

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#933 - Decay scheduler based on validation score

Issue - State: closed - Opened by cccs-km over 2 years ago - 4 comments

#932 - BPE support for transformer model

Issue - State: closed - Opened by nashid over 2 years ago - 1 comment
Labels: question

#931 - Comparison of OpenNMT-tf and tensor2tensor

Issue - State: closed - Opened by nashid over 2 years ago - 3 comments
Labels: question

#930 - Why is the SelfAttentionEncoderlayer inconsistent with the paper?

Issue - State: closed - Opened by Wenjun-Peng over 2 years ago - 2 comments

#929 - Add learning rate schedule InvSqrtDecay

Pull Request - State: closed - Opened by guillaumekln over 2 years ago

#928 - After tflite invoke, the tflite model use a larger memory.

Issue - State: closed - Opened by SefaZeng over 2 years ago - 5 comments

#927 - RuntimeError: Model diverged with loss = NaN.

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

#926 - Fine tuning BERT models for NLP tasks

Issue - State: open - Opened by cservan almost 3 years ago - 1 comment
Labels: enhancement

#925 - Allow setting all learning rate schedules parameters

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#924 - Support TensorFlow 2.8

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#923 - RsqrtDecay fails when decay_params scale is configured

Issue - State: closed - Opened by Dagamies almost 3 years ago - 2 comments

#921 - Update black to stable release

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#920 - Fix unknowns replacement when the source has BOS or EOS tokens

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#919 - Handle multi source length constraints in base Transformer config

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#917 - Import error on mac M1

Issue - State: closed - Opened by vietph34 almost 3 years ago - 1 comment

#916 - Is there a way to use dynamic decoding rather than fix output size for the tflite model?

Issue - State: open - Opened by SefaZeng almost 3 years ago - 1 comment
Labels: enhancement

#915 - Fail to load dynamic quantized tflite model

Issue - State: closed - Opened by SefaZeng almost 3 years ago - 5 comments

#914 - Declare alignment layers and heads when converting to CTranslate2

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#913 - Replace custom implementation with tf.nn.gelu

Pull Request - State: closed - Opened by guillaumekln almost 3 years ago

#912 - Translator inference is repeating the word and including values when the .

Issue - State: closed - Opened by eduamf almost 3 years ago - 1 comment