Add Typed Decisions benchmark eval results
#3
by SaylorTwift HF Staff - opened
.eval_results/autotrust__GEV-26B-Decide.yaml
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
- dataset:
|
| 2 |
+
id: LocalLLaMA/typed-decisions
|
| 3 |
+
task_id: accuracy
|
| 4 |
+
value: 0.522
|
| 5 |
+
date: "2026-10-16"
|
| 6 |
+
source:
|
| 7 |
+
url: https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals
|
| 8 |
+
name: "Typed Decisions benchmark run 2026-10-16 (results + leaderboard.md in bucket)"
|
| 9 |
+
notes: "Evaluated one question per request (the model's native /v1/decide interface), not the benchmark's single-request protocol; option/rubric descriptions embedded in the question text."
|
| 10 |
+
- dataset:
|
| 11 |
+
id: LocalLLaMA/typed-decisions
|
| 12 |
+
task_id: kl_from_gold
|
| 13 |
+
value: 0.4227
|
| 14 |
+
date: "2026-10-16"
|
| 15 |
+
source:
|
| 16 |
+
url: https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals
|
| 17 |
+
name: "Typed Decisions benchmark run 2026-10-16 (results + leaderboard.md in bucket)"
|
| 18 |
+
notes: "Evaluated one question per request (the model's native /v1/decide interface), not the benchmark's single-request protocol; option/rubric descriptions embedded in the question text."
|
| 19 |
+
- dataset:
|
| 20 |
+
id: LocalLLaMA/typed-decisions
|
| 21 |
+
task_id: brier
|
| 22 |
+
value: 0.2146
|
| 23 |
+
date: "2026-10-16"
|
| 24 |
+
source:
|
| 25 |
+
url: https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals
|
| 26 |
+
name: "Typed Decisions benchmark run 2026-10-16 (results + leaderboard.md in bucket)"
|
| 27 |
+
notes: "Evaluated one question per request (the model's native /v1/decide interface), not the benchmark's single-request protocol; option/rubric descriptions embedded in the question text."
|
| 28 |
+
- dataset:
|
| 29 |
+
id: LocalLLaMA/typed-decisions
|
| 30 |
+
task_id: ece
|
| 31 |
+
value: 0.0658
|
| 32 |
+
date: "2026-10-16"
|
| 33 |
+
source:
|
| 34 |
+
url: https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals
|
| 35 |
+
name: "Typed Decisions benchmark run 2026-10-16 (results + leaderboard.md in bucket)"
|
| 36 |
+
notes: "Evaluated one question per request (the model's native /v1/decide interface), not the benchmark's single-request protocol; option/rubric descriptions embedded in the question text."
|