Commits · c4447dabdf60467bdc8903029c25b0e2de428c99 · gaoqiong / lm-evaluation-harness

08 May, 2023 1 commit
- 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
05 Nov, 2021 1 commit
- Merge TokenizedLM and TorchLM into BaseLM · fea936e6
  Leo Gao authored Nov 05, 2021
  
  fea936e6
11 Oct, 2021 1 commit
- Refactor LM organization for more reuse · 7f24a08b
  Leo Gao authored Oct 11, 2021
  
  7f24a08b
06 Oct, 2021 1 commit
- Make gpt2 a little more tokenizer agnostic · e5066c69
  Leo Gao authored Oct 05, 2021
  
  e5066c69
29 Sep, 2021 2 commits
- add support for HF revision · b569f483
  Leo Gao authored Sep 29, 2021
  
  b569f483
- gpt2: fix bad multi gpu behavior and typecheck inputs · 7f7b717d
  Leo Gao authored Sep 28, 2021
  
  7f7b717d
22 May, 2021 1 commit

make create_from_arg_string fully general · 9d6d291f

Leo Gao authored May 21, 2021

for some reason putting it in LM and having it be inherited breaks
everything. should try to figure this out at some point.

9d6d291f

11 May, 2021 3 commits
- Update gpt2.py · ebe9e6b9
  sdtblck authored May 11, 2021
  
  ebe9e6b9
- gpt2: extract model call · e6decf02
  Leo Gao authored May 11, 2021
  
  e6decf02
- Overhaul command flags a bit · 5f42f976
  Leo Gao authored May 10, 2021
```
model_args should only be things that affect output of the model
therefore, stuff like batch size, device, etc shouldn't be in there
```
  5f42f976
05 May, 2021 1 commit
- Minor changes · f16c301e
  Leo Gao authored May 05, 2021
  
  f16c301e