Commits · bc103ce2b24ccd2001ed45375199ac1cb3567f81 · gaoqiong / lm-evaluation-harness

22 May, 2023 1 commit
- pad out all batches loglikelihood_rolling. confirmed correctness · 71388a7e
  Benjamin Fattori authored May 22, 2023
  
  71388a7e
20 May, 2023 3 commits
- unwrap gpt2 model to get n_ctx · c10c08a2
  Benjamin Fattori authored May 20, 2023
  
  c10c08a2
- improved warning message for processes < total gpus · e4f1dfb6
  Benjamin Fattori authored May 20, 2023
  
  e4f1dfb6
- unwrap model to get pos embs · 65127577
  Benjamin Fattori authored May 20, 2023
  
  65127577
19 May, 2023 1 commit
- pre-commit stuff · c4c20ff5
  lintangsutawika authored May 19, 2023
  
  c4c20ff5
16 May, 2023 1 commit
- added logging process · 8d608117
  lintangsutawika authored May 16, 2023
  
  8d608117
10 May, 2023 3 commits
- balance documents among hosts for wikitext · 8f859cd2
  Benjamin Fattori authored May 10, 2023
  
  8f859cd2
- drop comments about pytorch DP · 3cfd23a1
  Benjamin Fattori authored May 10, 2023
  
  3cfd23a1
- fix warning issue if single device · 4d3ea67a
  Benjamin Fattori authored May 10, 2023
  
  4d3ea67a
08 May, 2023 2 commits
- warning if launched incorrectly on multiGPU machine · 46f37ec0
  Benjamin Fattori authored May 08, 2023
  
  46f37ec0
- clean up cumul batches for rolling + remove unused print · c4447dab
  Benjamin Fattori authored May 08, 2023
  
  c4447dab
06 May, 2023 3 commits
- partial support for document-level padding · 32d7593d
  Benjamin Fattori authored May 06, 2023
  
  32d7593d
- assert + help message if script called incorrectly with multigpu · a8d9a449
  Benjamin Fattori authored May 06, 2023
  
  a8d9a449
- record _rank/_world_size for single GPU · a7d6cb2d
  Benjamin Fattori authored May 06, 2023
  
  a7d6cb2d
05 May, 2023 3 commits
- add utility func for slicing iterators, only tqdm on main process · eddd627a
  Benjamin Fattori authored May 05, 2023
  
  eddd627a
- bugfixes missed from local branch · 629bcfba
  Benjamin Fattori authored May 05, 2023
  
  629bcfba
- sync working changes with upstream · d4c5315a
  Benjamin Fattori authored May 05, 2023
  
  d4c5315a
02 May, 2023 1 commit
- prelim. multiple choice support · 487f7811
  haileyschoelkopf authored May 02, 2023
  
  487f7811
24 Apr, 2023 2 commits
- add template_vars + make jinja rendering stricter · 460584ca
  haileyschoelkopf authored Apr 23, 2023
  
  460584ca
- make tasks and models registered by decorators · f275301a
  haileyschoelkopf authored Apr 23, 2023
  
  f275301a
19 Apr, 2023 1 commit
- in-place replace main with lm-eval2, keeping old git history · d2a9b759
  haileyschoelkopf authored Apr 19, 2023
  
  d2a9b759
07 Mar, 2023 1 commit
- enable "low_cpu_mem_usage" · 859f96fd
  Xingjian Shi authored Mar 07, 2023
  
  859f96fd
04 Feb, 2023 1 commit
- hotfix(gpt2): Remove vocab-size logits slice · 1a545447
  jon-tow authored Feb 04, 2023
  
  1a545447
15 Sep, 2022 3 commits
- Remove stale tokenizer comment · 3dc27bd2
  jon-tow authored Sep 15, 2022
  
  3dc27bd2
- Move `TODO`s into one source location · 5b1beceb
  jon-tow authored Sep 15, 2022
  
  5b1beceb
- Manually concat tokenizer revision with subfolder · ff83e9b2
  jon-tow authored Sep 15, 2022
  
  ff83e9b2
03 May, 2022 2 commits
- fix codespell · d95a4333
  Fabrizio Milo authored May 02, 2022
  
  d95a4333
- add pre-commit · 121b7096
  Fabrizio Milo authored May 02, 2022
  
  121b7096
29 Apr, 2022 2 commits
- Revert "Merge branch 'master' into master" · 7585ec56
  jon-tow authored Apr 29, 2022
```
This reverts commit 5fe7e2c0, reversing
changes made to 14b8ec7e.
```
  7585ec56
- Fix max generation limit · eda365f6
  jon-tow authored Apr 28, 2022
  
  eda365f6
28 Apr, 2022 2 commits
- Modified stopping criteria for T5 and GPT-2 · 5e59320b
  Tian Yun authored Apr 28, 2022
  
  5e59320b
- Fixed generation trunction for GPT-2 and T5 · 6c86eb47
  Tian Yun authored Apr 28, 2022
  
  6c86eb47
27 Apr, 2022 1 commit
- Fixing stopping criteria · ecde9d2e
  Tian Yun authored Apr 27, 2022
  
  ecde9d2e
26 Apr, 2022 4 commits
- Modified stopping criteria for gpt2 · 61b57292
  Tian Yun authored Apr 26, 2022
  
  61b57292
- Added stoppping criteria for generation · 96ea7ddc
  Tian Yun authored Apr 26, 2022
  
  96ea7ddc
- Rename task specific to · f39c27c2
  cjlovering authored Apr 26, 2022
  
  f39c27c2
- Update the token encode · c26f1d4c
  cjlovering authored Apr 25, 2022
  
  c26f1d4c
30 Jan, 2022 1 commit
- Small fixes · 341663a9
  researcher2 authored Jan 31, 2022
  
  341663a9
26 Dec, 2021 1 commit
- Implement decontamination of evals against training set 13 grams. · 6597c347
  researcher2 authored Dec 25, 2021
  
  6597c347
24 Nov, 2021 1 commit
- Some refactor/clean-up, and GPT-3 update · 76ebb792
  Jason Phang authored Nov 23, 2021
  
  76ebb792