Commits · 21d897db56e8bf72bb0a2bdb3ace62cc7cba0bee · gaoqiong / lm-evaluation-harness

26 Apr, 2022 27 commits
- Updated the requests so that its easier to understand. · 21d897db
  cjlovering authored Apr 26, 2022
  
  21d897db
- Updated doc: If the answer choices is empty, then it is generation; else... · 4f85bcf9
  cjlovering authored Apr 26, 2022
```
Updated doc: If the answer choices is empty, then it is generation; else ranked choice. This will be the canonical approach when using PS.
```
  4f85bcf9
- Added t0, t5, mt5 · 716c87d6
  Tian Yun authored Apr 26, 2022
  
  716c87d6
- Add multi-reference ROUGE support · 94218002
  jon-tow authored Apr 26, 2022
  
  94218002
- A dependency required this but it was not installed by default · 9384ec91
  cjlovering authored Apr 26, 2022
  
  9384ec91
- Changing stopping criteria for coqa · 23a6ee7f
  cjlovering authored Apr 26, 2022
  
  23a6ee7f
- 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 · b2838b8d
  cjlovering authored Apr 26, 2022
  
  b2838b8d
- Minor updates to documentation. · 1dcca55c
  cjlovering authored Apr 26, 2022
  
  1dcca55c
- Added default behavior for bleu to the promtsourcetask class · d4c00093
  cjlovering authored Apr 26, 2022
  
  d4c00093
- Rename task specific to · f39c27c2
  cjlovering authored Apr 26, 2022
  
  f39c27c2
- Force no caching while testing · c27e29e1
  jon-tow authored Apr 26, 2022
  
  c27e29e1
- Update the token encode · c26f1d4c
  cjlovering authored Apr 25, 2022
  
  c26f1d4c
- Fix `wsc` subset name · b988137d
  jon-tow authored Apr 25, 2022
  
  b988137d
- Print table across task and prompt · 422380cc
  cjlovering authored Apr 25, 2022
  
  422380cc
- Turned off generation tasks for now. Changed process to look at the metrics.... · 0cdcc989
  jon-tow authored Apr 25, 2022
```
Turned off generation tasks for now. Changed process to look at the metrics. Only accuracy implemented.
```
  0cdcc989
- Add `HANS` · 0e578306
  jon-tow authored Apr 25, 2022
  
  0e578306
- QQP with PS integration. · b1a3c6e3
  cjlovering authored Apr 25, 2022
  
  b1a3c6e3
- MRPC with PS integration. · 26e94211
  cjlovering authored Apr 25, 2022
  
  26e94211
- QNLI with PS integration. · b44aa554
  cjlovering authored Apr 25, 2022
  
  b44aa554
- SST with PS integration. (It was already done.) · 4c201b97
  cjlovering authored Apr 25, 2022
  
  4c201b97
- SST with PS integration. (It was already done.) · e49cf8da
  cjlovering authored Apr 25, 2022
  
  e49cf8da
- Temp. commenting our e2e · 31a019c2
  cjlovering authored Apr 25, 2022
  
  31a019c2
- Removed the default option for an acc task · c93093b6
  cjlovering authored Apr 25, 2022
  
  c93093b6
- Add `higher_is_better` & `aggregation` defaults to `PromptSourceTask` · 4ae2ab37
  jon-tow authored Apr 25, 2022
  
  4ae2ab37
- Add `eos_token` property · 6ec93da2
  jon-tow authored Apr 25, 2022
  
  6ec93da2
25 Apr, 2022 6 commits
- Add multiple tasks · 34f591af
  jon-tow authored Apr 25, 2022
  
  34f591af
- Fix prompt source rank choice accuracy · 2bfa4518
  jon-tow authored Apr 25, 2022
  
  2bfa4518
- Fix task name to template creation · 9f388461
  jon-tow authored Apr 25, 2022
  
  9f388461
- Fix coqa · 9484eecc
  jon-tow authored Apr 25, 2022
  
  9484eecc
- Add PromptSourceTask to the updated tasks · 7d282b5f
  cjlovering authored Apr 25, 2022
  
  7d282b5f
- Initial integration · 88745155
  cjlovering authored Apr 25, 2022
  
  88745155
31 Mar, 2022 3 commits
- Fix docstring · 9434722c
  Jon Tow authored Mar 31, 2022
  
  9434722c
- Adopt `_process_doc` method · 39c8c87f
  Jon Tow authored Mar 31, 2022
  
  39c8c87f
- Add specific `load_dataset` args · d79a4389
  Jon Tow authored Mar 31, 2022
  
  d79a4389
19 Mar, 2022 1 commit
- Whitespace formatting · 4f8b99f6
  Jon Tow authored Mar 19, 2022
  
  4f8b99f6
18 Mar, 2022 3 commits
- Fix dataset value feature · c2e32987
  Jon Tow authored Mar 18, 2022
  
  c2e32987
- Add custom `image`less `HeadQA` dataset · 897ae76d
  Jon Tow authored Mar 18, 2022
  
  897ae76d
- Fix training docs caching · 13317f8c
  Jon Tow authored Mar 18, 2022
  
  13317f8c