PerryLink commited on
Commit ·
a570303
1
Parent(s): 5eb9395
eval-results: drop the stale dataset revision pin so the board can list us
Browse filesAll four entries pinned LocalLLaMA/typed-decisions at e039ebff..., which was the
dataset tip on 2026-10-08 and is now behind 7189f46b.... Empirically, every model
currently on the auto-populated board omits this field, and every model that pins a
now-stale revision - ours among them - is absent from it. The field is optional per
the Hub's own eval-results spec, so removing it loses nothing: the value, date,
source and disclosure notes are unchanged, and accuracy still reports 0.791.
.eval_results/typed-decisions.yaml
CHANGED
|
@@ -8,11 +8,17 @@
|
|
| 8 |
# board reads each model repo's own eval results, so that entry is up to the
|
| 9 |
# submitters") is in the dataset discussion #25.
|
| 10 |
# After push, verify: GET https://huggingface.co/api/datasets/LocalLLaMA/typed-decisions/leaderboard
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
- dataset:
|
| 13 |
id: LocalLLaMA/typed-decisions
|
| 14 |
task_id: accuracy
|
| 15 |
-
revision: e039ebffcc280174dd354227424fb2b249f191de
|
| 16 |
value: 0.791
|
| 17 |
date: '2026-10-15'
|
| 18 |
source:
|
|
@@ -42,7 +48,6 @@
|
|
| 42 |
- dataset:
|
| 43 |
id: LocalLLaMA/typed-decisions
|
| 44 |
task_id: kl_from_gold
|
| 45 |
-
revision: e039ebffcc280174dd354227424fb2b249f191de
|
| 46 |
value: 0.0548
|
| 47 |
date: '2026-10-09'
|
| 48 |
source:
|
|
@@ -57,7 +62,6 @@
|
|
| 57 |
- dataset:
|
| 58 |
id: LocalLLaMA/typed-decisions
|
| 59 |
task_id: brier
|
| 60 |
-
revision: e039ebffcc280174dd354227424fb2b249f191de
|
| 61 |
value: 0.0248
|
| 62 |
date: '2026-10-09'
|
| 63 |
source:
|
|
@@ -72,7 +76,6 @@
|
|
| 72 |
- dataset:
|
| 73 |
id: LocalLLaMA/typed-decisions
|
| 74 |
task_id: ece
|
| 75 |
-
revision: e039ebffcc280174dd354227424fb2b249f191de
|
| 76 |
value: 0.2519
|
| 77 |
date: '2026-10-09'
|
| 78 |
source:
|
|
|
|
| 8 |
# board reads each model repo's own eval results, so that entry is up to the
|
| 9 |
# submitters") is in the dataset discussion #25.
|
| 10 |
# After push, verify: GET https://huggingface.co/api/datasets/LocalLLaMA/typed-decisions/leaderboard
|
| 11 |
+
#
|
| 12 |
+
# 2026-10-15 (second change): the `revision:` pin was removed from all four entries.
|
| 13 |
+
# It pinned the dataset at e039ebffcc280174dd354227424fb2b249f191de, which was the
|
| 14 |
+
# dataset tip on 2026-10-08 but is now stale (tip is 7189f46b...). Every model that
|
| 15 |
+
# currently appears on the auto-populated board omits this field; the four that pin a
|
| 16 |
+
# now-stale revision - ours among them - do not appear at all. The rule is empirical,
|
| 17 |
+
# not documented, so it is recorded here rather than presented as spec.
|
| 18 |
|
| 19 |
- dataset:
|
| 20 |
id: LocalLLaMA/typed-decisions
|
| 21 |
task_id: accuracy
|
|
|
|
| 22 |
value: 0.791
|
| 23 |
date: '2026-10-15'
|
| 24 |
source:
|
|
|
|
| 48 |
- dataset:
|
| 49 |
id: LocalLLaMA/typed-decisions
|
| 50 |
task_id: kl_from_gold
|
|
|
|
| 51 |
value: 0.0548
|
| 52 |
date: '2026-10-09'
|
| 53 |
source:
|
|
|
|
| 62 |
- dataset:
|
| 63 |
id: LocalLLaMA/typed-decisions
|
| 64 |
task_id: brier
|
|
|
|
| 65 |
value: 0.0248
|
| 66 |
date: '2026-10-09'
|
| 67 |
source:
|
|
|
|
| 76 |
- dataset:
|
| 77 |
id: LocalLLaMA/typed-decisions
|
| 78 |
task_id: ece
|
|
|
|
| 79 |
value: 0.2519
|
| 80 |
date: '2026-10-09'
|
| 81 |
source:
|