microsoft/fastseq issues and pull requests

#122 - Fix Axure pipeline for unit tests

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

#121 - Running error with PyTorch 1.12.1

Issue - State: closed - Opened by StevenTang1998 over 2 years ago - 2 comments

#120 - EL-attention GPT-2

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

#119 - In which file to read the source code implementation of El-Attention for self-attention

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

#118 - Where to read EL-Attention source code for huggingface-transformers

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

#117 - Avoid torch/extension.h in cu file

Pull Request - State: closed - Opened by feihugis over 2 years ago - 1 comment

#116 - gpt2 attention bug fix

Pull Request - State: closed - Opened by JulianneKnott almost 3 years ago - 2 comments

#115 - Can the fastseq install on windows?

Issue - State: closed - Opened by vivid-k almost 3 years ago - 3 comments

#114 - Add tokenizer options

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

#113 - Support outputting multiple sequences for each sample in hf models.

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

#112 - Update hf baseline to remove repetition

Pull Request - State: closed - Opened by JulianneKnott almost 3 years ago - 1 comment

#111 - additional parameters for hf

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

#110 - small bug fix / more flexible hf cli

Pull Request - State: closed - Opened by JulianneKnott about 3 years ago

#107 - Transformers v4.12.0 compatible

Pull Request - State: closed - Opened by JulianneKnott about 3 years ago

#106 - NMT models speedup abnormally related to batch size

Issue - State: open - Opened by dearchill over 3 years ago - 2 comments

#105 - Does fastseq support cpu

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

#104 - Fairseq v0.10.2 compatible

Pull Request - State: closed - Opened by JulianneKnott over 3 years ago - 6 comments

#103 - Edit cite

Pull Request - State: closed - Opened by yuyan2do over 3 years ago

#102 - Add 'Open in VS Code' badge

Pull Request - State: closed - Opened by feihugis over 3 years ago

#101 - Fix the parser issue in supporting prophetnet task

Pull Request - State: closed - Opened by feihugis over 3 years ago

#100 - Fix Colab link

Pull Request - State: closed - Opened by yuyan2do over 3 years ago

#99 - Support for current fairseq 0.10.2

Issue - State: closed - Opened by alaneckhardt over 3 years ago - 13 comments

#98 - Support for Hugging Face's PEGASUS Model

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

#95 - Support for HF's transformers 3.1+

Issue - State: open - Opened by tingofurro over 3 years ago - 6 comments

#94 - Support for ONNX models & INT8 quantization

Issue - State: open - Opened by bkaruman over 3 years ago - 2 comments

#93 - Does it support model seq2seq with encoder, and decoder base on lstm, bi-lstm?

Issue - State: closed - Opened by trangtv57 over 3 years ago - 2 comments

#92 - Any end-end inference example with Google Colab & HuggingFace

Issue - State: open - Opened by ramsrigouthamg over 3 years ago - 2 comments

#91 - doc: minor typo

Pull Request - State: closed - Opened by monologg over 3 years ago

#90 - Update README.md spelling

Pull Request - State: closed - Opened by cep21 over 3 years ago

#89 - Does it support Tensorflow 2?

Issue - State: closed - Opened by s4sarath over 3 years ago - 3 comments

#88 - Revise README and add doc for El attention

Pull Request - State: closed - Opened by yuyan2do over 3 years ago - 2 comments

#87 - fairseq eval_lm

Issue - State: closed - Opened by sshleifer almost 4 years ago - 2 comments

#86 - fix unilm token type, parameters & tests issues

Pull Request - State: closed - Opened by fuliucansheng almost 4 years ago - 1 comment

#85 - Including El Attn optimization

Pull Request - State: closed - Opened by NickNickGo almost 4 years ago

#84 - Update README

Pull Request - State: closed - Opened by feihugis almost 4 years ago

#83 - Utf-8 encode for source and target text.

Pull Request - State: closed - Opened by JiushengChen about 4 years ago - 1 comment

#82 - Enable fastseq generate CLI to handle the empty shard correctly

Pull Request - State: closed - Opened by feihugis about 4 years ago

#81 - Optimize GPT2

Pull Request - State: closed - Opened by feihugis about 4 years ago - 7 comments

#80 - Unilm Model Support & GenerationMixinV3

Pull Request - State: closed - Opened by fuliucansheng about 4 years ago - 11 comments

#79 - Early exit for ngram blocking.

Pull Request - State: closed - Opened by JiushengChen about 4 years ago

#78 - ModuleNotFoundError: No module named 'fastseq.models'

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

#74 - Benchmark refinement

Pull Request - State: closed - Opened by JiushengChen about 4 years ago

#73 - Add scripts for preparing binary dataset for ProphetNet

Pull Request - State: closed - Opened by feihugis about 4 years ago

#72 - Clean and fix benchmarks.

Pull Request - State: closed - Opened by JiushengChen about 4 years ago

#71 - exception guarding cuda extension import

Pull Request - State: open - Opened by NickNickGo about 4 years ago

#70 - RuntimeError: CUDA error: no kernel image is available for execution on the device

Issue - State: closed - Opened by sshleifer about 4 years ago - 7 comments

#69 - Detokenization parallelization public repo

Pull Request - State: closed - Opened by NickNickGo about 4 years ago - 13 comments

#68 - Update benchmark for bart

Pull Request - State: closed - Opened by yuyan2do about 4 years ago - 3 comments

#66 - Sort and unsort the input texts for transformers

Pull Request - State: open - Opened by feihugis about 4 years ago - 2 comments

#65 - Update wmt benchmark.

Pull Request - State: closed - Opened by JiushengChen about 4 years ago

#64 - Update the instructions for installation

Pull Request - State: closed - Opened by feihugis about 4 years ago

#63 - Script for building manylinux wheel

Pull Request - State: open - Opened by feihugis about 4 years ago

#62 - Add missing init files

Pull Request - State: closed - Opened by feihugis about 4 years ago

#61 - made ngram op device agnostic, unit test cleaned

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 1 comment

#60 - ACTION REQUIRED: Microsoft needs this private repository to complete compliance info

Issue - State: closed - Opened by microsoft-github-operations[bot] over 4 years ago

#59 - Update install_requires and enable fairseq to work with torch 1.6&1.7

Pull Request - State: closed - Opened by feihugis over 4 years ago

#58 - Fix prophenet dict loading.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago

#57 - Update speed in benchmark

Pull Request - State: closed - Opened by JiushengChen over 4 years ago - 7 comments

#56 - Generate the XML log file for each fastseq unit test

Pull Request - State: closed - Opened by feihugis over 4 years ago - 4 comments

#55 - Speedup hs distilbart download.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago - 2 comments

#54 - [WIP] Support TorchScript and graph rewrite

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

#53 - Add READMEs for models.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago

#52 - updating benchmarks of latest repo

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 7 comments

#51 - Remove some trailing spaces

Pull Request - State: closed - Opened by feihugis over 4 years ago - 1 comment

#50 - minor change in CI pipeline

Pull Request - State: closed - Opened by NickNickGo over 4 years ago

#49 - PLEASE IGNORE , dummy PR for azure CI pipeline test

Pull Request - State: closed - Opened by NickNickGo over 4 years ago

#48 - azure pipeline spec

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 1 comment

#47 - Azure ci pipeline

Pull Request - State: closed - Opened by NickNickGo over 4 years ago

#46 - Decode hypothesis in Fairseq v0.9.0.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago - 1 comment

#45 - Add the missing init.py for models module

Pull Request - State: closed - Opened by feihugis over 4 years ago - 1 comment

#44 - Target at only one major version of transformers + fairseq in the main branch

Pull Request - State: closed - Opened by feihugis over 4 years ago - 2 comments

#43 - Encoder-decoder Multihead attention cpu optimization

Pull Request - State: open - Opened by NickNickGo over 4 years ago - 1 comment

#42 - Update docs

Pull Request - State: closed - Opened by feihugis over 4 years ago

#41 - Fix transformers failures

Pull Request - State: closed - Opened by feihugis over 4 years ago

#40 - Cuda op for ngram repeat blocking

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 2 comments

#39 - Transformers unit tests failure

Issue - State: closed - Opened by NickNickGo over 4 years ago

#38 - Enhance pylint rcfile

Pull Request - State: closed - Opened by feihugis over 4 years ago

#37 - detokenization parallelization

Pull Request - State: open - Opened by NickNickGo over 4 years ago - 6 comments

#36 - Update README.md

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 2 comments

#35 - Fix the error of dic changed size during iteration

Pull Request - State: closed - Opened by feihugis over 4 years ago

#34 - Add Prophetnet

Pull Request - State: closed - Opened by feihugis over 4 years ago - 7 comments

#33 - Guarding user env with test wrappers

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 7 comments

#32 - fairseq/transformers unit test modify local environment

Issue - State: closed - Opened by JiushengChen over 4 years ago - 3 comments

#31 - Add the configs for cache and log envs

Pull Request - State: closed - Opened by feihugis over 4 years ago

#30 - Update the speed range for T5 benchmark

Pull Request - State: closed - Opened by feihugis over 4 years ago

#29 - Fix the fairseq test

Pull Request - State: closed - Opened by feihugis over 4 years ago

#28 - Errors in test_fairseq_optimizer.py

Issue - State: closed - Opened by JiushengChen over 4 years ago - 1 comment

#27 - Fix an issue in replace decorator

Pull Request - State: closed - Opened by feihugis over 4 years ago

#26 - memory not-release issue for large BS on FastSeq

Issue - State: closed - Opened by yetingqiaqia over 4 years ago - 3 comments

#25 - Refactor logging and resolve some coding style issues

Pull Request - State: closed - Opened by feihugis over 4 years ago - 1 comment

#24 - Fix return status in summary.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago

#23 - T5 speed

Issue - State: closed - Opened by JiushengChen over 4 years ago - 6 comments

#22 - Fix a typo in hf_distibart.sh

Pull Request - State: closed - Opened by feihugis over 4 years ago

#21 - Fix wmt benchmark.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago - 1 comment

#20 - Added parallelization , speed bump in decoding approx 1.83

Pull Request - State: closed - Opened by NickNickGo over 4 years ago - 9 comments

#19 - Break down benchmarks by models. Add fairseq lm model.

Pull Request - State: closed - Opened by JiushengChen over 4 years ago

#18 - Simplify ngram block algo

Pull Request - State: closed - Opened by yuyan2do over 4 years ago - 7 comments

#17 - filter data before copy and switch to torch.multiprocessing

Pull Request - State: closed - Opened by yuyan2do over 4 years ago

#16 - Support BART model in Transformers-2.11.0

Pull Request - State: closed - Opened by feihugis over 4 years ago - 2 comments

#15 - Optimize Transformers-T5

Pull Request - State: closed - Opened by feihugis over 4 years ago - 4 comments

GitHub / microsoft/fastseq issues and pull requests