s3prl/s3prl issues and pull requests

#462 - HuBERT as ER downstream task

Issue - State: open - Opened by celsofranssa about 2 years ago - 1 comment

#460 - Add "device" to not overwrite args (Update run_downstream.py)

Pull Request - State: closed - Opened by pabloriera about 2 years ago - 1 comment

#459 - Add "device" to cannot_overwrite_args

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

#458 - About SD result in paper

Issue - State: open - Opened by liyunlongaaa about 2 years ago

#457 - s3prl upstream key issue with local checkpoint

Issue - State: open - Opened by bloodraven66 about 2 years ago

#456 - oodasr

Pull Request - State: closed - Opened by dannyshicar about 2 years ago - 1 comment

#455 - Fix BYOL-A implementation

Pull Request - State: closed - Opened by daisukelab about 2 years ago - 1 comment

#454 - Is there no vq_apc local in s3prl?

Issue - State: open - Opened by kaen2891 about 2 years ago

#453 - SID task loss function.

Issue - State: open - Opened by raotnameh about 2 years ago - 1 comment

#452 - Bump setuptools from 59.5.0 to 65.5.1 in /requirements

Pull Request - State: closed - Opened by dependabot[bot] about 2 years ago
Labels: dependencies

#451 - unspecifed upstream models

Issue - State: closed - Opened by marziye-A about 2 years ago - 3 comments

#450 - requirements does not have tensorboard

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

#449 - ContentVec support

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

#448 - Add fast sv testing

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

#447 - Update Flashlight Text/Sequence repo sources

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

#446 - Why don't we have a ASR and PR task for libri-other? Just curious.

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

#445 - support hf and fairseq wav2vec2/hubert

Pull Request - State: closed - Opened by leo19941227 about 2 years ago

#444 - distill hubert GPU Memory-Usage is so big

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

#443 - Mandarin Hubert

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

#442 - Commonvoice version 7.0 is not easily accessible.

Issue - State: closed - Opened by raotnameh about 2 years ago - 5 comments

#441 - Add multilingual upstream (#434)

Pull Request - State: closed - Opened by leo19941227 about 2 years ago

#440 - add `wav2vec2_base_ll60k`

Pull Request - State: closed - Opened by vectominist about 2 years ago - 1 comment

#439 - Add new wav2vec2 entries: s2st-es-voxpopuli

Pull Request - State: closed - Opened by andi611 over 2 years ago - 1 comment

#438 - Requests download

Pull Request - State: closed - Opened by eric891224 over 2 years ago - 1 comment

#437 - Use requests package for stabler download

Issue - State: closed - Opened by leo19941227 over 2 years ago - 1 comment

#436 - Support entry for loading fairseq checkpoint directly if fairseq is installed.

Issue - State: closed - Opened by leo19941227 over 2 years ago - 1 comment
Labels: enhancement, good second issue

#435 - Add SSL models with unsupervised ASR

Issue - State: open - Opened by ftshijt over 2 years ago - 1 comment
Labels: enhancement

#434 - Add variants of HuBERT and wav2vec2

Issue - State: closed - Opened by ftshijt over 2 years ago - 5 comments
Labels: enhancement

#433 - test

Issue - State: closed - Opened by leo19941227 over 2 years ago

#432 - Reproducing the IC downstream task

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

#431 - ASV downstream training is painfully slow. IS there anything can be done to speed it up?

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

#430 - Bad performance on real audio

Issue - State: open - Opened by benam2 over 2 years ago - 11 comments

#429 - Passt

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#428 - Ensure audio models

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#427 - ASR configuration file has wrong default value

Pull Request - State: closed - Opened by Comet0322 over 2 years ago - 1 comment

#424 - Error running IC intent classification

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

#423 - fix doc for importing Featurizer

Pull Request - State: closed - Opened by leo19941227 over 2 years ago - 1 comment

#422 - Featurizer has some bugs

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

#421 - remove speechbrain dataset dependency for sound processing tasks

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#420 - MTWV metric

Issue - State: closed - Opened by fatemeh1291374 over 2 years ago - 3 comments

#419 - [Feature Request] A prediction pipeline for combining upstream and downstream model

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

#418 - Auto resample hear

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#417 - How to use custom in-house model?

Issue - State: open - Opened by CBrun38 over 2 years ago - 1 comment

#416 - No such file or directory while opening './4-gram.arpa.gz' KenLM

Issue - State: closed - Opened by benam2 over 2 years ago - 6 comments

#415 - Different results when batching waveforms

Issue - State: open - Opened by pabloriera over 2 years ago

#414 - Make sure the results align with the old codebase

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#413 - fix doc toc

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#412 - NameError: name 'CriterionType' is not defined in W2lDecoder

Issue - State: closed - Opened by benam2 over 2 years ago

#410 - What is the role of the downsample_rate variable？

Issue - State: closed - Opened by YangHao97 over 2 years ago - 3 comments

#409 - Change wavlm unispeech link

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#408 - Ecapa

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

#407 - Flashlight Decoder --> Flashlight Text

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

#405 - add ecapa-tdnn

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

#404 - refine dataset and batch sampler building process

Pull Request - State: closed - Opened by leo19941227 over 2 years ago

#402 - There are tasks that ESPNET does with S3PRL that fail

Issue - State: closed - Opened by abcdbosh over 2 years ago - 18 comments
Labels: enhancement

#395 - Incorrect encoding detected in ./downstream/asr/char.dict, please rebuild the dataset

Issue - State: closed - Opened by benam2 over 2 years ago - 9 comments

#393 - (WIP) a better version of enhancement and separation downstream

Pull Request - State: closed - Opened by HuangZiliAndy over 2 years ago - 12 comments

#391 - Added the save and load features code.

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

#390 - It would be better to save the features if upstream is not trainable

Issue - State: open - Opened by raotnameh over 2 years ago - 3 comments
Labels: enhancement

#389 - TypeError: 'builtin_function_or_method' object does not support item assignment

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

#388 - Last Hidden Layers Embedding for Emotion Model

Issue - State: closed - Opened by VSK02101996 over 2 years ago - 12 comments

#385 - How to convert a superb pretrained model to a fairseq model?

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

#383 - SUPERB Leaderboard Artefacts

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

#382 - Inference for Source Separation

Issue - State: open - Opened by bryant0918 over 2 years ago - 3 comments

#381 - KeyError: 'unix_user' when pip install -e ./

Issue - State: open - Opened by rocketsfallonrocketfalls over 2 years ago - 1 comment

#375 - error: Segmentation fault (core dumped)

Issue - State: closed - Opened by haha010508 over 2 years ago - 3 comments

#374 - weird error of DDP train when use self.best_score

Issue - State: open - Opened by leijue222 over 2 years ago - 6 comments

#372 - loss function matching between the paper and the project implementation

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

#368 - More detailed usage of SV tasks

Issue - State: closed - Opened by leijue222 over 2 years ago - 2 comments

#367 - Pypi installation

Issue - State: closed - Opened by franp9am over 2 years ago - 2 comments

#361 - Use of downstream tasks of WavLM ASV Testing

Issue - State: closed - Opened by leijue222 over 2 years ago - 1 comment

#360 - SUPERB BNECHMARK.

Issue - State: open - Opened by raotnameh over 2 years ago - 19 comments
Labels: FAQ, paper / research / method

#354 - Tensorboard question

Issue - State: closed - Opened by gyccc000 over 2 years ago - 2 comments

#343 - prepare upstream for downstream (WIP, for #341)

Pull Request - State: open - Opened by leo19941227 over 2 years ago - 12 comments

#341 - BYOL-A is incompletely implemented...

Issue - State: open - Opened by daisukelab over 2 years ago - 10 comments
Labels: enhancement

#340 - Downstream ASR Custom dataset

Issue - State: closed - Opened by ana-kuznetsova almost 3 years ago - 2 comments
Labels: customize

#338 - DDP train falied

Issue - State: open - Opened by TCL606 almost 3 years ago - 3 comments

#283 - Downstream ASV: training terminated

Issue - State: open - Opened by YangHao97 about 3 years ago - 7 comments
Labels: environment related

#281 - Error in mos_prediction (for finding file `system_mos_all_trackwise.csv`)

Issue - State: closed - Opened by bagustris about 3 years ago - 7 comments

#280 - Multi-gpu is taking more time than single gpu

Issue - State: closed - Opened by Sindhu-Hegde about 3 years ago - 6 comments

#271 - Which wav2vec2 model should I choose for reproducing SUPERB?

Issue - State: closed - Opened by YangHao97 about 3 years ago - 6 comments

#268 - What are these training configurations of downstream tasks in SUPERB paper?

Issue - State: closed - Opened by mechanicalsea about 3 years ago - 5 comments

GitHub / s3prl/s3prl issues and pull requests