Vveerrgg commited on
Commit
c5caeac
·
verified ·
1 Parent(s): d43e176

card: feedback and security section

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -200,7 +200,7 @@ and per-row source credits) is published so the results can be checked:
200
  not to recall them. For a specific historical dish, give it the text.
201
  - The judges are models. Four open judges from four families agree on the direction and size of the changes, and
202
  we keep checking them against people with Label Test, our blind labelling tool. **Help refine it:** label five
203
- answers at [label-test.humanjava.com](https://label-test.humanjava.com/?s=cook4&k=cook-9ac741a1e387) (no account; we keep only a name or initials and your labels).
204
  - Still invents on some no-recipe questions: about 1 in 40 when three of four judges must agree, about 1 in 16 by the
205
  strictest single judge. Most remaining cases are borderline hedges (a common default offered with a caveat).
206
  - Strongest in English and Spanish; Thai, Tamil and Hindi had under 200 source records each.
@@ -211,7 +211,7 @@ and per-row source credits) is published so the results can be checked:
211
  - **The model got something wrong** (invented a detail, misread a recipe, unsafe advice): please
212
  [open a report](https://github.com/HumanjavaEnterprises/qwen3-8b-cooking-assistant/issues/new?template=wrong-answer.yml).
213
  Real examples are what improve the next version.
214
- - **Help measure it:** [label five answers](https://label-test.humanjava.com/?s=cook4&k=cook-9ac741a1e387) on Label Test.
215
  - **Security** (files that don't match `SHA256SUMS`, unexpected embedded content): report privately, see
216
  [SECURITY.md](https://github.com/HumanjavaEnterprises/qwen3-8b-cooking-assistant/blob/main/SECURITY.md).
217
 
 
200
  not to recall them. For a specific historical dish, give it the text.
201
  - The judges are models. Four open judges from four families agree on the direction and size of the changes, and
202
  we keep checking them against people with Label Test, our blind labelling tool. **Help refine it:** label five
203
+ answers at [label-test.humanjava.com](https://label-test.humanjava.com/?s=cook4&k=cook-9ac741a1e387&src=hf-card) (no account; we keep only a name or initials and your labels).
204
  - Still invents on some no-recipe questions: about 1 in 40 when three of four judges must agree, about 1 in 16 by the
205
  strictest single judge. Most remaining cases are borderline hedges (a common default offered with a caveat).
206
  - Strongest in English and Spanish; Thai, Tamil and Hindi had under 200 source records each.
 
211
  - **The model got something wrong** (invented a detail, misread a recipe, unsafe advice): please
212
  [open a report](https://github.com/HumanjavaEnterprises/qwen3-8b-cooking-assistant/issues/new?template=wrong-answer.yml).
213
  Real examples are what improve the next version.
214
+ - **Help measure it:** [label five answers](https://label-test.humanjava.com/?s=cook4&k=cook-9ac741a1e387&src=hf-card) on Label Test.
215
  - **Security** (files that don't match `SHA256SUMS`, unexpected embedded content): report privately, see
216
  [SECURITY.md](https://github.com/HumanjavaEnterprises/qwen3-8b-cooking-assistant/blob/main/SECURITY.md).
217