Code

sync#11

Code

Merged

spatil6 merged 185 commits intospatil6:tf_ctrl_seq_classificationspatil6/transformers:tf_ctrl_seq_classificationfrom huggingface:masterhuggingface/transformers:masterCopy head branch name to clipboardDec 16, 2020

+40,866-52,416

Commits

Commits on Nov 22, 2020

• Fix many typos (#8708)

  bryant1410

  authored
  e1f3156

  Copy full SHA for e1f3156

Commits on Nov 23, 2020

• Create README.md (#8630)

  Show description for b6d864e
  

  moniquebm

  and

  julien-c

  authored
  b6d864e

  Copy full SHA for b6d864e

• added bangla-bert-sentiment model card (#8687)

  sagorbrur

  authored
  b5187e3

  Copy full SHA for b5187e3

• create README.md (#8682)

  Show description for 52585e4
  

  bino282

  and

  julien-c

  authored
  52585e4

  Copy full SHA for 52585e4

• [model_cards] Add card for gpt2-rnm (#8673)

  e-tornike

  authored
  48cc224

  Copy full SHA for 48cc224

• Fix bug in x-attentions output for roberta and harden test to catch it (#8660)

  

  Yossi Synett

  authored
  18c8cf0

  Copy full SHA for 18c8cf0

• Add pip install update to resolve import error in transformers notebook (#8616)

  Show description for 143b564
  

  jessicayung

  and

  LysandreJik

  authored
  143b564

  Copy full SHA for 143b564

• [model_cards]: control input examples of Geotrend models (#8727)

  Show description for eec7661
  

  Amine Abdaoui

  authored
  eec7661

  Copy full SHA for eec7661

• Improve bert-japanese tokenizer handling (#8659)

  Show description for 0cc5ab1
  

  julien-c

  and

  sgugger

  authored
  0cc5ab1

  Copy full SHA for 0cc5ab1

• Change default cache path (#8734)

  Show description for 9000242
  

  sgugger

  and

  LysandreJik

  authored
  9000242

  Copy full SHA for 9000242

• [trainer] make generate work with multigpu (#8716)

  Show description for 1e45bef
  stas00

  authored
  1e45bef

  Copy full SHA for 1e45bef

• gpt2 and t5 parallel modeling (#8696)

  Show description for 1cd9be2
  

  3 peopleauthored
  1cd9be2

  Copy full SHA for 1cd9be2

• Document new training argument

  sgugger

  committed
  49759c0

  Copy full SHA for 49759c0

• consistent ignore keys + make private (#8737)

  Show description for e84786a
  stas00

  authored
  e84786a

  Copy full SHA for e84786a

• Fix max length in run_plm script (#8738)

  sgugger

  authored
  367f497

  Copy full SHA for 367f497

• Add early stopping callback to pytorch trainer (#8581)

  Show description for 8ffc01a
  

  Colin Brochtrup

  authored
  8ffc01a

  Copy full SHA for 8ffc01a

• Update TF BERT test

  LysandreJik

  committed
  e1b7e10

  Copy full SHA for e1b7e10

• TF BERT test update

  LysandreJik

  committed
  7f2c009

  Copy full SHA for 7f2c009

Commits on Nov 24, 2020

• Model parallel documentation (#8741)

  Show description for 02f48b9
  LysandreJik

  authored
  02f48b9

  Copy full SHA for 02f48b9

• [EsperBERTo] Fix URLs to assets

  julien-c

  committed
  9e71aa2

  Copy full SHA for 9e71aa2

• Support various BERT relative position embeddings (2nd) (#8276)

  Show description for 2c83b3c
  

  zhiheng-huang

  and

  patrickvonplaten

  authored
  2c83b3c

  Copy full SHA for 2c83b3c

• Fix slow tests v2 (#8746)

  Show description for 6fdd0bb
  LysandreJik

  authored
  6fdd0bb

  Copy full SHA for 6fdd0bb

• MT5 should have an autotokenizer (#8743)

  Show description for e09e54f
  LysandreJik

  authored
  e09e54f

  Copy full SHA for e09e54f

• added instructions for syncing upstream master with forked master via PR (#8745)

  Show description for 8d4ed7e
  

  bdalal

  and

  stas00

  authored
  8d4ed7e

  Copy full SHA for 8d4ed7e

• fix rag index names in eval_rag.py example (#8730)

  lhoestq

  authored
  a7d73cf

  Copy full SHA for a7d73cf

• [core] implement support for run-time dependency version checking (#8645)

  Show description for 82d443a
  

  stas00

  and

  sgugger

  authored
  82d443a

  Copy full SHA for 82d443a

• New TF model inputs (#8602)

  Show description for 29d4992
  jplu

  authored
  29d4992

  Copy full SHA for 29d4992

• Create README.md (#8761)

  mrm8488

  authored
  90d5ab3

  Copy full SHA for 90d5ab3

Commits on Nov 25, 2020

• Big model table (#8774)

  Show description for 4821ea5
  

  sgugger

  and

  julien-c

  authored
  4821ea5

  Copy full SHA for 4821ea5

• Fix QA argument handler (#8765)

  Show description for 138f45c
  

  LysandreJik

  and

  Narsil

  authored
  138f45c

  Copy full SHA for 138f45c

• Return correct Bart hidden state tensors (#8747)

  Show description for 369f1d7
  

  joeddav

  and

  patrickvonplaten

  authored
  369f1d7

  Copy full SHA for 369f1d7

• [XLNet] Fix mems behavior (#8567)

  Show description for 2a6fbe6
  patrickvonplaten

  authored
  2a6fbe6

  Copy full SHA for 2a6fbe6

• Create README.md (#8752)

  

  joangines

  authored
  30e7f7e

  Copy full SHA for 30e7f7e

Commits on Nov 26, 2020

• finetune.py: specifying generation min_length (#8478)

  

  Daniel Khashabi

  authored
  5aa361f

  Copy full SHA for 5aa361f

• Create README.md (#8760)

  mrm8488

  authored
  66e9608

  Copy full SHA for 66e9608

• Revert "finetune.py: specifying generation min_length (#8478)" (#8805)

  Show description for 8f07f5c
  patrickvonplaten

  authored
  8f07f5c

  Copy full SHA for 8f07f5c

• Fix PPLM (#8779)

  Show description for 52708d2
  

  chutaklee

  and

  patrickvonplaten

  authored
  52708d2

  Copy full SHA for 52708d2

• potpurri of small fixes (#8807)

  stas00

  authored
  ddf3c64

  Copy full SHA for ddf3c64

• typo (#8810)

  stas00

  authored
  cb7602b

  Copy full SHA for cb7602b

Commits on Nov 27, 2020

• [FlaxBert] Fix non-broadcastable attention mask for batched forward-passes (#8791)

  Show description for f8eda59
  KristianHolsheimer

  authored
  f8eda59

  Copy full SHA for f8eda59

• Update README.md (#8815)

  Show description for e3ef62b
  mdermentzi

  authored
  e3ef62b

  Copy full SHA for e3ef62b

• [Flax test] Add require pytorch to flix flax test (#8816)

  Show description for a2cf375
  patrickvonplaten

  authored
  a2cf375

  Copy full SHA for a2cf375

• Fix dpr<>bart config for RAG (#8808)

  Show description for a7d46a0
  patrickvonplaten

  authored
  a7d46a0

  Copy full SHA for a7d46a0

• Extend typing to path-like objects in PretrainedConfig and PreTrainedModel (#8770)

  Show description for f9a2a9e
  gcompagnoni

  authored
  f9a2a9e

  Copy full SHA for f9a2a9e

• Create README.md (#8729)

  Show description for 03bddc3
  mrm8488

  authored
  03bddc3

  Copy full SHA for 03bddc3

• Fix setup.py (#8798)

  Show description for b0f2dbc
  jplu

  authored
  b0f2dbc

  Copy full SHA for b0f2dbc

• Add barthez model (#8393)

  Show description for 81fe0bf
  

  moussaKam

  and

  sgugger

  authored
  81fe0bf

  Copy full SHA for 81fe0bf

• BART & FSMT: fix decoder not returning hidden states from the last layer (#8597)

  Show description for 0a921b6
  

  MaksymDel

  and

  stas00

  authored
  0a921b6

  Copy full SHA for 0a921b6

• suggest a numerical limit of 50MB for determining @slow (#8824)

  stas00

  authored
  00ea456

  Copy full SHA for 00ea456

• Temporarily deactivate model generation

  LysandreJik

  committed
  edbff1f

  Copy full SHA for edbff1f

• Model parallel tests should return, not pass in non model parallel settings. (#8825)

  LysandreJik

  authored
  18c32ee

  Copy full SHA for 18c32ee

Commits on Nov 28, 2020

• fix mt5 config (#8832)

  patrickvonplaten

  authored
  36b60ce

  Copy full SHA for 36b60ce

Commits on Nov 29, 2020

• [Pegasus] Refactor Tokenizer (#8731)

  Show description for 5ced23d
  patrickvonplaten

  authored
  5ced23d

  Copy full SHA for 5ced23d

• Minor docs typo fixes (#8797)

  Show description for 3a08cc1
  

  guyrosin

  and

  guyrosin

  authored
  3a08cc1

  Copy full SHA for 3a08cc1

• [CI] implement job skipping for doc-only PRs (#8826)

  Show description for c239dcd
  stas00

  authored
  c239dcd

  Copy full SHA for c239dcd

Commits on Nov 30, 2020

• Migration guide from v3.x to v4.x (#8763)

  Show description for 610cb10
  

  LysandreJik

  and

  sgugger

  authored
  610cb10

  Copy full SHA for 610cb10

• Add T5 Encoder for Feature Extraction (#8717)

  Show description for 40ecaf0
  

  3 peopleauthored
  40ecaf0

  Copy full SHA for 40ecaf0

• token-classification: use is_world_process_zero instead of deprecated is_world_master() (#8828)

  stefan-it

  authored
  19fa01c

  Copy full SHA for 19fa01c

• Correct docstring. (#8845)

  Show description for cc983cd
  Fraser-Greenlee

  authored
  cc983cd

  Copy full SHA for cc983cd

• Add a direct link to the big table (#8850)

  sgugger

  authored
  75f8100

  Copy full SHA for 75f8100

• Comment the skip job on doc line

  sgugger

  committed
  08e7076

  Copy full SHA for 08e7076

• Merge remote-tracking branch 'origin/master'

  sgugger

  committed
  4062c75

  Copy full SHA for 4062c75

• Use model.from_pretrained for DataParallel also (#8795)

  Show description for 7738494
  shaie

  authored
  7738494

  Copy full SHA for 7738494

• Remove deprecated evalutate_during_training (#8852)

  Show description for 5530299
  

  sgugger

  and

  LysandreJik

  authored
  5530299

  Copy full SHA for 5530299

• Release: v4.0.0

  LysandreJik

  committed
  22b0ff7

  Copy full SHA for 22b0ff7

• Update docs

  LysandreJik

  committed
  9995a34

  Copy full SHA for 9995a34

• Attempt to fix Flax CI error(s) (#8829)

  Show description for 51b0713
  

  mfuntowicz

  and

  sgugger

  authored
  51b0713

  Copy full SHA for 51b0713

• fix pypi complaint on version naming

  LysandreJik

  committed
  5fd3d81

  Copy full SHA for 5fd3d81

• NerPipeline (TokenClassification) now outputs offsets of words (#8781)

  Show description for d8fc26e
  Narsil

  authored
  d8fc26e

  Copy full SHA for d8fc26e

• [s2s trainer] fix DP mode (#8823)

  Show description for 7f34d75
  stas00

  authored
  7f34d75

  Copy full SHA for 7f34d75

Commits on Dec 1, 2020

• Ctrl for sequence classification (#8812)

  Show description for 4a9e502
  elk-cloner

  authored
  4a9e502

  Copy full SHA for 4a9e502

• Fix doc for language code (#8848)

  RQuispeC

  authored
  814b955

  Copy full SHA for 814b955

• 2 typos in modeling_rag.py (#8676)

  Show description for d366228
  ratthachat

  authored
  d366228

  Copy full SHA for d366228

• Make the big table creation/check platform independent (#8856)

  sgugger

  authored
  c0df963

  Copy full SHA for c0df963

• Prevent BatchEncoding from blindly passing casts down to the tensors it contains. Fixes #6582. (#8860)

  Show description for 9c18f15
  

  Craigacp

  and

  LysandreJik

  authored
  9c18f15

  Copy full SHA for 9c18f15

• Better warning when loading a tokenizer with AutoTokenizer w/o SnetencePiece (#8881)

  LysandreJik

  authored
  a947386

  Copy full SHA for a947386

• [CI] skip docs-only jobs take #2 (#8853)

  Show description for 21db560
  stas00

  authored
  21db560

  Copy full SHA for 21db560

• Better support for resuming training (#8878)

  sgugger

  authored
  7c10dd2

  Copy full SHA for 7c10dd2

• Add a parallel_mode property to TrainingArguments (#8877)

  Show description for b08843c
  sgugger

  authored
  b08843c

  Copy full SHA for b08843c

• start using training_args.parallel_mode (#8882)

  stas00

  authored
  379005c

  Copy full SHA for 379005c

• disable job skip - need more work

  Show description for 693ac35
  stas00

  authored
  693ac35

  Copy full SHA for 693ac35

Commits on Dec 2, 2020

• [ci] skip doc jobs take #3 (#8885)

  Show description for 24f0c2f
  stas00

  authored
  24f0c2f

  Copy full SHA for 24f0c2f

• Transfoxl seq classification (#8868)

  Show description for f6b44e6
  spatil6

  authored
  f6b44e6

  Copy full SHA for f6b44e6

• Warning about too long input for fast tokenizers too (#8799)

  Show description for a8c3f9a
  Narsil

  authored
  a8c3f9a

  Copy full SHA for a8c3f9a

• [trainer] improve code readability (#8903)

  Show description for 7e1cb00
  stas00

  authored
  7e1cb00

  Copy full SHA for 7e1cb00

• Fix typo in docstring (#8905)

  ryota-mo

  authored
  801b2cb

  Copy full SHA for 801b2cb

• Update README.md (#8906)

  devangi2000

  authored
  e52f9c0

  Copy full SHA for e52f9c0

• [PyTorch] Refactor Resize Token Embeddings (#8880)

  Show description for 443f67e
  patrickvonplaten

  authored
  443f67e

  Copy full SHA for 443f67e

Commits on Dec 3, 2020

• [model_cards] lm-head was deprecated

  Show description for 2b7fc9a
  julien-c

  committed
  2b7fc9a

  Copy full SHA for 2b7fc9a

• Don't warn that models aren't available if Flax is available. (#8841)

  skye

  authored
  0deece9

  Copy full SHA for 0deece9

• Avoid erasing the attention mask when double padding (#8915)

  sgugger

  authored
  8453201

  Copy full SHA for 8453201

• Fix move when the two cache folders exist (#8917)

  sgugger

  authored
  6ed7e32

  Copy full SHA for 6ed7e32

• Tweak wording + Add badge w/ number of models on the hub (#8914)

  Show description for 9ad6194
  

  julien-c

  and

  sgugger

  authored
  9ad6194

  Copy full SHA for 9ad6194

• Put Transformers on Conda (#8918)

  Show description for 0c5615a
  

  LysandreJik

  and

  sgugger

  authored
  0c5615a

  Copy full SHA for 0c5615a

• Patch model parallel test (#8920)

  Show description for aa60b23
  LysandreJik

  authored
  aa60b23

  Copy full SHA for aa60b23

Commits on Dec 4, 2020

• [s2s finetune_trainer] add instructions for distributed training (#8884)

  stas00

  authored
  4c3d98d

  Copy full SHA for 4c3d98d

• Better booleans handling in the TF models (#8777)

  Show description for dcd3046
  jplu

  authored
  dcd3046

  Copy full SHA for dcd3046

• Fix TF T5 only encoder model with booleans (#8925)

  LysandreJik

  authored
  71688a8

  Copy full SHA for 71688a8

• [ci] skip doc jobs - circleCI is not reliable - disable skip for now (#8926)

  Show description for 73c51f7
  stas00

  authored
  73c51f7

  Copy full SHA for 73c51f7

• [seq2seq] document the caveat of leaky native amp (#8930)

  Show description for df311a5
  

  stas00

  and

  sgugger

  authored
  df311a5

  Copy full SHA for df311a5

Commits on Dec 5, 2020

• Don't pass in token_type_ids to BART for GLUE (#8929)

  Show description for 8dfc8c7
  ethanjperez

  authored
  8dfc8c7

  Copy full SHA for 8dfc8c7

• Fix typo for modeling_bert import resulting in ImportError (#8931)

  Show description for ef93a25
  machelreid

  authored
  ef93a25

  Copy full SHA for ef93a25

Commits on Dec 6, 2020

• Add model card (#8948)

  Show description for 72d6c9c
  

  sarnikowski

  and

  julien-c

  authored
  72d6c9c

  Copy full SHA for 72d6c9c

Commits on Dec 7, 2020

• Fix QA pipeline on Windows (#8947)

  sgugger

  authored
  28c77dd

  Copy full SHA for 28c77dd

• Add TFGPT2ForSequenceClassification based on DialogRPT (#8714)

  Show description for 483e132
  spatil6

  authored
  483e132

  Copy full SHA for 483e132

• Remove sourcerer (#8965)

  clmnt

  authored
  de6befd

  Copy full SHA for de6befd

• Use word_ids to get labels in run_ner (#8962)

  Show description for 7f9ccff
  sgugger

  authored
  7f9ccff

  Copy full SHA for 7f9ccff

• > 30 files leads to hanging on --More--

  Show description for 37f4c24
  stas00

  authored
  37f4c24

  Copy full SHA for 37f4c24

• Update README.txt (#8957)

  bino282

  authored
  7ccd973

  Copy full SHA for 7ccd973

• Create README.md (#8964)

  wietsedv

  authored
  0bce7c5

  Copy full SHA for 0bce7c5

• transformers-cli: LFS multipart uploads (> 5GB) (#8663)

  Show description for 28fa014
  

  4 peopleauthored
  28fa014

  Copy full SHA for 28fa014

• Small fix to the run clm script (#8973)

  sgugger

  authored
  62d30e0

  Copy full SHA for 62d30e0

• add max_length to showcase the use of truncation (#8975)

  navjotts

  authored
  c108d0b

  Copy full SHA for c108d0b

• Copyright (#8970)

  Show description for 00aa9db
  sgugger

  authored
  00aa9db

  Copy full SHA for 00aa9db

Commits on Dec 8, 2020

• Check table as independent script (#8976)

  LysandreJik

  authored
  2ae7388

  Copy full SHA for 2ae7388

• [training] SAVE_STATE_WARNING was removed in pytorch (#8979)

  Show description for 9d7d000
  stas00

  authored
  9d7d000

  Copy full SHA for 9d7d000

• Optional layers (#8961)

  Show description for bf7f79c
  jplu

  authored
  bf7f79c

  Copy full SHA for bf7f79c

• [model_card] remove bogus testing changes

  julien-c

  authored
  0d9e6ca

  Copy full SHA for 0d9e6ca

• Make ModelOutput pickle-able (#8989)

  sgugger

  authored
  04c446f

  Copy full SHA for 04c446f

• Fix interaction of return_token_type_ids and add_special_tokens (#8854)

  LysandreJik

  authored
  b7cdd00

  Copy full SHA for b7cdd00

• Removed unused encoder_hidden_states and encoder_attention_mask (#8972)

  Show description for 7809eb8
  guillaume-be

  authored
  7809eb8

  Copy full SHA for 7809eb8

• New squad example (#8992)

  Show description for 447808c
  

  sgugger

  and

  patrickvonplaten

  authored
  447808c

  Copy full SHA for 447808c

• Templates overhaul 1 (#8993)

  LysandreJik

  authored
  67ff1c3

  Copy full SHA for 67ff1c3

Commits on Dec 9, 2020

• Diverse beam search 2 (#9006)

  Show description for 02d0e03
  

  3 peopleauthored
  02d0e03

  Copy full SHA for 02d0e03

• Fix link to stable version in the doc navbar (#9007)

  sgugger

  authored
  7e1d709

  Copy full SHA for 7e1d709

• Remove use of deprected method in Trainer HP search (#8996)

  sgugger

  authored
  61abd50

  Copy full SHA for 61abd50

• push (#9008)

  patrickvonplaten

  authored
  da37a21

  Copy full SHA for da37a21

• Add the code_search_net dataset tag to CodeBERTa model cards (#9005)

  SBrandeis

  authored
  e977ed2

  Copy full SHA for e977ed2

• fixes #8968 (#9009)

  cronoik

  authored
  8729109

  Copy full SHA for 8729109

• Add MP Net 2 (#9004)

  

  StillKeepTry

  authored and

  LysandreJik

  committed
  df2af6d

  Copy full SHA for df2af6d

• Flax Masked Language Modeling training example (#8728)

  Show description for 7562714
  

  mfuntowicz

  and

  marcvanzee

  authored
  7562714

  Copy full SHA for 7562714

• [Bart] Refactor - fix issues, consistency with the library, naming (#8900)

  Show description for 06971ac
  patrickvonplaten

  authored
  06971ac

  Copy full SHA for 06971ac

• [wip] [ci] doc-job-skip take #4 dry-run (#8980)

  Show description for 5e637e6
  stas00

  authored
  5e637e6

  Copy full SHA for 5e637e6

Commits on Dec 10, 2020

• ✏️ Fix typo (#9020)

  astariul

  authored
  c95de29

  Copy full SHA for c95de29

• Fix documention of book in LayoutLM (#9017)

  sgugger

  authored
  35bffd7

  Copy full SHA for 35bffd7

• MPNet copyright files (#9015)

  sgugger

  authored
  51e81e5

  Copy full SHA for 51e81e5

• Enforce all objects in the main init are documented (#9014)

  sgugger

  authored
  1310e1a

  Copy full SHA for 1310e1a

• Refactor FLAX tests (#9034)

  sgugger

  authored
  8d4bb02

  Copy full SHA for 8d4bb02

• Fix typo #9012 (#1) (#9038)

  Show description for 91ab02a
  NatLun137

  authored
  91ab02a

  Copy full SHA for 91ab02a

• Remove value error (#8985)

  Show description for b01ddc9
  jplu

  authored
  b01ddc9

  Copy full SHA for b01ddc9

Commits on Dec 11, 2020

• 🎨 Change nn.dropout to layer.Dropout (#9047)

  astariul

  authored
  935e346

  Copy full SHA for 935e346

• Create README.md (#9030)

  Show description for 5e794b6
  panggi

  authored
  5e794b6

  Copy full SHA for 5e794b6

• Initial README for t5-base-indonesian-summarization-cased model (#9028)

  Show description for 649d389
  panggi

  authored
  649d389

  Copy full SHA for 649d389

• Update README.md (#8820)

  moniquebm

  authored
  b161f1a

  Copy full SHA for b161f1a

• QARiB Arabic and dialects models (#8796)

  Show description for 76df559
  

  3 peopleauthored
  76df559

  Copy full SHA for 76df559

• Create README.md (#8751)

  Show description for c615df7
  

  joangines

  and

  julien-c

  authored
  c615df7

  Copy full SHA for c615df7

• Create README.md (#8281)

  Show description for 5527f78
  

  RamonMamon

  and

  julien-c

  authored
  5527f78

  Copy full SHA for 5527f78

• Create README.md (#8096)

  Show description for 7c8f5f6
  

  ganeshkharad2

  and

  julien-c

  authored
  7c8f5f6

  Copy full SHA for 7c8f5f6

• update tatoeba workflow (#9051)

  patil-suraj

  authored
  86896de

  Copy full SHA for 86896de

• Reorganize examples (#9010)

  Show description for 783d7d2
  

  3 peopleauthored
  783d7d2

  Copy full SHA for 783d7d2

• Fix PreTrainedTokenizer.pad when first inputs are empty (#9018)

  Show description for 70527ba
  sgugger

  authored
  70527ba

  Copy full SHA for 70527ba

• Remove docs only check (#9065)

  LysandreJik

  authored
  91fa707

  Copy full SHA for 91fa707

• Bump notebook in /examples/research_projects/movement-pruning/lxmert (#9062)

  Show description for 24f6cde
  dependabot[bot]

  authored
  24f6cde

  Copy full SHA for 24f6cde

• Make ProphetNetModel really compatible with EncoderDecoder (#9033)

  Show description for 9cc9f41
  patrickvonplaten

  authored
  9cc9f41

  Copy full SHA for 9cc9f41

• Fix min_null_pred in the run_qa script (#9067)

  sgugger

  authored
  29e4597

  Copy full SHA for 29e4597

• [model_cards] Migrate cards from this repo to model repos on huggingface.co (#9013)

  Show description for 3552d0e
  

  3 peopleauthored
  3552d0e

  Copy full SHA for 3552d0e

Commits on Dec 14, 2020

• Fix embeddings resizing in TF models (#8657)

  Show description for 51d9c56
  jplu

  authored
  51d9c56

  Copy full SHA for 51d9c56

• Patch *ForCausalLM model (#9092)

  LysandreJik

  authored
  6587cf9

  Copy full SHA for 6587cf9

• [RAG, Bart] Align RAG, Bart cache with T5 and other models of transformers (#9098)

  Show description for fa1ddce
  patrickvonplaten

  authored
  fa1ddce

  Copy full SHA for fa1ddce

• correct var name in TrainingArguments docstring (#9096)

  navjotts

  authored
  d6af344

  Copy full SHA for d6af344

• Fixed a broken link in documentation (#9101)

  SBrandeis

  authored
  74daf1f

  Copy full SHA for 74daf1f

• Testing Experimental CI Features (#9070)

  stas00

  authored
  b00eb4f

  Copy full SHA for b00eb4f

• Add parallelization support for T5EncoderModel (#9082)

  Show description for a9c8bff
  

  agemagician

  and

  Ahmed Elnaggar

  authored
  a9c8bff

  Copy full SHA for a9c8bff

• Fix T5 and BART for TF (#9063)

  Show description for df3f4d2
  jplu

  authored
  df3f4d2

  Copy full SHA for df3f4d2

• Pin TF to < 2.4

  sgugger

  committed
  e4ef57a

  Copy full SHA for e4ef57a

• Also pin TF CPU

  sgugger

  committed
  251eb70

  Copy full SHA for 251eb70

Commits on Dec 15, 2020

• [finetune_trainer] enhancements and fixes (#9042)

  Show description for c19d046
  

  3 peopleauthored
  c19d046

  Copy full SHA for c19d046

• fix a bug in eval_batch_retrieval (#9089)

  yoshitomo-matsubara

  authored
  44c340f

  Copy full SHA for 44c340f

• Clarify use of TrainingArguments.disable_tqdm in Jupyter Notebooks (#9076)

  Show description for ed1845e
  

  lewtun

  and

  sgugger

  authored
  ed1845e

  Copy full SHA for ed1845e

• native amp leak fix landed in 1.7.1 (#9115)

  Show description for 14c79c3
  stas00

  authored
  14c79c3

  Copy full SHA for 14c79c3

• Fix stack overflow (#9114)

  LysandreJik

  authored
  59da3f2

  Copy full SHA for 59da3f2

• Fix T5 model parallel tes (#9107)

  Show description for 6ccea04
  LysandreJik

  authored
  6ccea04

  Copy full SHA for 6ccea04

• Fix tf2.4 (#9120)

  Show description for ef2d4cd
  

  jplu

  and

  sgugger

  authored
  ef2d4cd

  Copy full SHA for ef2d4cd

• Added TF OpenAi GPT1 Sequence Classification (#9105)

  Show description for 389aba3
  

  spatil6

  and

  LysandreJik

  authored
  389aba3

  Copy full SHA for 389aba3

• [TF Bart] Refactor TFBart (#9029)

  Show description for abc573f
  patrickvonplaten

  authored
  abc573f

  Copy full SHA for abc573f

• Fix typo in trainer_tf.py (#9132)

  luckynozomi

  authored
  3caba8d

  Copy full SHA for 3caba8d

• fix bart loss masking (#9131)

  patrickvonplaten

  authored
  80bdb9c

  Copy full SHA for 80bdb9c

• correct mistake in order (#9134)

  patrickvonplaten

  authored
  d018622

  Copy full SHA for d018622

• Fix Bart Shift (#9135)

  Show description for 18ecd36
  patrickvonplaten

  authored
  18ecd36

  Copy full SHA for 18ecd36

• Fix add order (#9129)

  jplu

  authored
  e771749

  Copy full SHA for e771749

• [Examples] Add automatic dataset splitting in language-modeling examples (#9133)

  Show description for 2a7e8e1
  TevenLeScao

  authored
  2a7e8e1

  Copy full SHA for 2a7e8e1

• Add large model config (#9140)

  LysandreJik

  authored
  0b2f46f

  Copy full SHA for 0b2f46f

• Add possibility to switch between APEX and AMP in Trainer (#9137)

  Show description for ad895af
  

  sgugger

  and

  stas00

  authored
  ad895af

  Copy full SHA for ad895af

• [WIP] Tapas v4 (tres) (#9117)

  Show description for 1551e2d
  

  4 peopleauthored
  1551e2d

  Copy full SHA for 1551e2d

• Fix fp16_backend field

  sgugger

  committed
  51adb97

  Copy full SHA for 51adb97