Datasets:
Tasks:
Text Classification
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
structured-decisions
calibration
probabilistic-classification
system-one
workflow-evaluation
Synthetic
License:
Add Decision Models rows (#38): quyet-1-large, s1-pro, s1-fast; regenerate chart
Browse files- README.md +38 -36
- assets/leaderboard.png +2 -2
README.md
CHANGED
|
@@ -69,39 +69,42 @@ the two tables are not comparable.
|
|
| 69 |
|
| 70 |
| # | Model | Kind | Accuracy ↑ | KL from gold ↓ | Brier ↓ | ECE ↓ | p50 latency | Price / 1M input |
|
| 71 |
|---|---|---|---|---|---|---|---|---|
|
| 72 |
-
| 1 | **[
|
| 73 |
-
| 2 | [
|
| 74 |
-
| 3 | [Amos 1
|
| 75 |
-
| 4 | [
|
| 76 |
-
| 5 | [
|
| 77 |
-
| 6 | [
|
| 78 |
-
| 7 | [
|
| 79 |
-
| 8 | [
|
| 80 |
-
| 9 | [
|
| 81 |
-
| 10 | [
|
| 82 |
-
| 11 | [
|
| 83 |
-
| 12 | [
|
| 84 |
-
| 13 |
|
| 85 |
-
| 14 | [
|
| 86 |
-
| 15 |
|
| 87 |
-
| 16 | [
|
| 88 |
-
| 17 | [
|
| 89 |
-
| 18 | [
|
| 90 |
-
| 19 | [
|
| 91 |
-
| 20 | [
|
| 92 |
-
| 21 | [
|
| 93 |
-
| 22 | [
|
| 94 |
-
| 23 | [
|
| 95 |
-
| 24 | [
|
| 96 |
-
| 25 | [
|
| 97 |
-
| 26 | [
|
| 98 |
-
| 27 | [d3-
|
| 99 |
-
| 28 | [
|
| 100 |
-
| 29 | [Jeff-
|
| 101 |
-
| 30 | [
|
| 102 |
-
| 31 | [
|
| 103 |
-
| 32 | [
|
| 104 |
-
| 33 | [
|
|
|
|
|
|
|
|
|
|
| 105 |
| – | Prior (ignores the input) | reference | 0.470 | 0.347 | 0.189 | 0.088 | – | – |
|
| 106 |
| – | Uniform (same probability on every option) | reference | 0.308 | 0.444 | 0.238 | 0.169 | – | – |
|
| 107 |
|
|
@@ -114,7 +117,7 @@ Hosted rows are p50 end to end from a client, one request at a time.
|
|
| 114 |
‡ Reported by the submitter on their own hardware (a different GPU and a different unit for each), not comparable with the other rows.
|
| 115 |
§ Self-reported by the submitter in the linked discussion and not re-run by us. Metric definitions follow the submitter's report; ECE in particular is computed differently by different submitters.
|
| 116 |
◆ Scored by a third party ([SaylorTwift](https://huggingface.co/SaylorTwift), discussion #37) on Hugging Face Jobs, 2026-10-16 to 10-19, with one public scorer for every row and not re-run by us. Raw predictions and scripts are in the [results bucket](https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals). Its accuracy counts a tie in the gold as correct and its ECE uses 10 bins, so small differences from the other rows can come from the scorer. Latency was measured on mixed Jobs hardware and is left out. d3-turbo was served through its own `system_one()` runtime because vLLM does not load it yet.
|
| 117 |
-
♠
|
| 118 |
|
| 119 |
### Fitted or fine-tuned on `train`
|
| 120 |
|
|
@@ -159,8 +162,7 @@ model is learning the teacher's quirks.
|
|
| 159 |
|
| 160 |
◇ Clean re-score by the submitter after removing 500 training pairs derived from model predictions on `test` rows ([discussion #25](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/25)). The originally submitted 0.906 was measured on a split the model had partly seen, so it is not listed; KL, Brier and ECE were only reported for that run. Do not train on `test` rows, `test` labels, or anything derived from predictions on them.
|
| 161 |
|
| 162 |
-
**[meraGPT Decider 1](https://meragpt.com/models/state-decider-1?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=typed-decisions) is
|
| 163 |
-
the art among zero-shot models on this benchmark.** It leads Jev on every question type (`noul` 0.840 vs
|
| 164 |
0.775, `choice` 0.733 vs 0.720, `score` 0.739 vs 0.696), its distributions sit far
|
| 165 |
closer to the gold (KL 0.096 vs 1.442), it is faster end to end, and it costs less
|
| 166 |
per token. It answers `POST /v1/systemone` at
|
|
|
|
| 69 |
|
| 70 |
| # | Model | Kind | Accuracy ↑ | KL from gold ↓ | Brier ↓ | ECE ↓ | p50 latency | Price / 1M input |
|
| 71 |
|---|---|---|---|---|---|---|---|---|
|
| 72 |
+
| 1 | **[Quyet-1.0-Large](https://huggingface.co/chinhnc/Quyet-1.0-Large)** (31B) via [Decision Models](https://decisionmodels.io) (`quyet-1-large`) ♠ ([#38](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/38)) | general, zero-shot, open weights | **0.791** | 0.281 | 0.114 | 0.093 | 3,335 ms♠ | $0.035 |
|
| 73 |
+
| 2 | [meraGPT Decider 1](https://meragpt.com/models/state-decider-1?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=typed-decisions) (`sd-1`) | general, zero-shot | 0.768 | **0.096** | **0.052** | 0.180 | **526 ms** | $0.03 |
|
| 74 |
+
| 3 | [Amos 1 medium](https://huggingface.co/mukaddamzaid/amos-1-medium) (9B) § ([#31](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/31)) | general, zero-shot, open weights | 0.7645 | 0.151 | 0.080 | 0.103 | 118 ms‡ (H100) | open weights |
|
| 75 |
+
| 4 | [Amos 1 large](https://huggingface.co/mukaddamzaid/amos-1-large) (27B) § ([#31](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/31)) | general, zero-shot, open weights | 0.761 | 0.126 | 0.069 | 0.097 | 357 ms‡ (H100) | open weights |
|
| 76 |
+
| 5 | [pplx-decider-v1.1-27b](https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b) (27B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.753 | 0.441 | 0.160 | 0.060 | – | open weights |
|
| 77 |
+
| 6 | [d3](https://huggingface.co/vllm-sr/d3) (27B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.749 | 0.409 | 0.146 | 0.049 | – | open weights |
|
| 78 |
+
| 7 | [Darwin-27B-ZTC](https://huggingface.co/FINAL-Bench/Darwin-27B-ZTC) (27B) § ([#17](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/17)) | general, zero-shot, open weights | 0.743 | 0.204 | 0.097 | – | – | open weights |
|
| 79 |
+
| 8 | [d3-mini](https://huggingface.co/vllm-sr/d3-mini) (4B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.743 | 0.319 | 0.123 | 0.047 | – | open weights |
|
| 80 |
+
| 9 | [Liquid AI d1](https://docs.liquid.ai/lfm/models/decision-models) (`d1:free`) | general, zero-shot | 0.742 | 0.475 | 0.155 | 0.124 | 525 ms | free tier |
|
| 81 |
+
| 10 | [linne-oak 1.0](https://huggingface.co/PNPM244/linne-oak-1-0) (31B) § ([#35](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/35)) | general, zero-shot, open weights | 0.739 | 0.709 | 0.191 | – | 1.3 s‡ (8 concurrent) | open weights |
|
| 82 |
+
| 11 | [vLLM-SR d3](https://huggingface.co/vllm-sr/d3) (26B, FP8) via [Decision Models](https://decisionmodels.io) (`s1-pro`) ♠ ([#38](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/38)) | general, zero-shot, open weights | 0.739 | 0.437 | 0.151 | 0.124 | 3,405 ms♠ | $0.08 |
|
| 83 |
+
| 12 | [d3-turbo](https://huggingface.co/vllm-sr/d3-turbo) (26B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.736 | 0.438 | 0.162 | 0.069 | – | open weights |
|
| 84 |
+
| 13 | [clef](https://huggingface.co/Cloudflare/clef) (27B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.728 | 0.191 | 0.103 | 0.025 | – | open weights |
|
| 85 |
+
| 14 | [openjev](https://huggingface.co/openjev/openjev) (27B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.728 | 0.506 | 0.191 | 0.101 | – | open weights |
|
| 86 |
+
| 15 | TypeSafe Jev 1.13.0 | general, zero-shot | 0.727 | 1.442 | 0.148 | 0.144 | 710 ms | $0.042 |
|
| 87 |
+
| 16 | [Blink v0.3](https://huggingface.co/PixilabAI/Blink-v0.3-26B-A4B-NVFP4) (26B-A4B, NVFP4) via [Decision Models](https://decisionmodels.io) (`s1-fast`) ♠ ([#38](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/38)) | general, zero-shot, open weights | 0.727 | 0.153 | 0.079 | 0.129 | 3,274 ms♠ | $0.019 |
|
| 88 |
+
| 17 | [d3-flash](https://huggingface.co/vllm-sr/d3-flash) (9B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.717 | 0.322 | 0.131 | 0.031 | – | open weights |
|
| 89 |
+
| 18 | [Featherless Simple Jev](https://simple-jev.featherless.ai) (`Qwen3.6-35B-A3B-classifier`) | general, zero-shot | 0.716 | 0.488 | 0.176 | – | – | free demo |
|
| 90 |
+
| 19 | [jev-vision-27b](https://huggingface.co/e12ex2/jev-vision-27b) (Qwen3.8-27B + LoRA) § | general, zero-shot, open weights | 0.714 | 0.318 | 0.129 | 0.026 | 199 ms‡ (H100) | open weights |
|
| 91 |
+
| 20 | [clef-flash](https://huggingface.co/Cloudflare/clef-flash) (9B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.710 | 0.209 | 0.111 | 0.016 | – | open weights |
|
| 92 |
+
| 21 | [prima-ratio + 12B](https://github.com/andrea-tomassi/prima-ratio) § | general, zero-shot | 0.702 | 0.564 | 0.234 | 0.146 | 700 ms‡ | local GGUF |
|
| 93 |
+
| 22 | [JEV-35B](https://huggingface.co/autotrust/JEV-35B) (35B MoE) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.700 | 0.526 | 0.169 | 0.062 | – | open weights |
|
| 94 |
+
| 23 | [JEV-27B-VL](https://huggingface.co/autotrust/JEV-27B-VL) (28B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.688 | 0.544 | 0.194 | 0.113 | – | open weights |
|
| 95 |
+
| 24 | [clef-omni](https://huggingface.co/Cloudflare/clef-omni) (35B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.673 | 0.247 | 0.123 | 0.030 | – | open weights |
|
| 96 |
+
| 25 | [OpenDecider-small](https://huggingface.co/manjunathshiva/opendecider-small) § | general, zero-shot, open weights | 0.671 | 0.211 | 0.117 | – | 40 ms‡ | open weights |
|
| 97 |
+
| 26 | [d1-3B](https://huggingface.co/LiquidAI/d1-3B) (3B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.656 | 0.286 | 0.144 | 0.037 | – | open weights |
|
| 98 |
+
| 27 | [d3-nano](https://huggingface.co/vllm-sr/d3-nano) (2B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.629 | 0.545 | 0.211 | 0.113 | – | open weights |
|
| 99 |
+
| 28 | [Bongard-mini](https://huggingface.co/AgentBull/bongard-mini) § | general, zero-shot, open weights | 0.594 | 0.256 | 0.132 | 0.067 | 225 ms‡ | open weights |
|
| 100 |
+
| 29 | [Jeff-Gemma4-E2B](https://huggingface.co/mstrasser/Jeff-Gemma4-E2B) | general, zero-shot, open weights | 0.561 | 0.403 | 0.219 | 0.188 | 2,272 ms† | open weights |
|
| 101 |
+
| 30 | [d3-lite](https://huggingface.co/vllm-sr/d3-lite) (0.8B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.531 | 0.518 | 0.244 | 0.130 | – | open weights |
|
| 102 |
+
| 31 | [GEV-26B-Decide](https://huggingface.co/autotrust/GEV-26B-Decide) (26B) ◆♠ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.522 | 0.423 | 0.215 | 0.066 | – | open weights |
|
| 103 |
+
| 32 | [Jeff-Qwen3.5-2B](https://huggingface.co/mstrasser/Jeff-Qwen3.5-2B) | general, zero-shot, open weights | 0.511 | 0.460 | 0.237 | 0.203 | 1,346 ms† | open weights |
|
| 104 |
+
| 33 | [JEV-27B](https://huggingface.co/autotrust/JEV-27B) (27B) ◆♠ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.495 | 0.550 | 0.253 | 0.136 | – | open weights |
|
| 105 |
+
| 34 | [Jeff-Qwen3.5-0.8B](https://huggingface.co/mstrasser/Jeff-Qwen3.5-0.8B) | general, zero-shot, open weights | 0.483 | 0.679 | 0.313 | 0.251 | 662 ms† | open weights |
|
| 106 |
+
| 35 | [d3-edge](https://huggingface.co/vllm-sr/d3-edge) (0.6B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.425 | 1.176 | 0.456 | 0.341 | – | open weights |
|
| 107 |
+
| 36 | [THX-01](https://huggingface.co/doofz/THX-01) (0.3B) ◆ ([#37](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/37)) | general, zero-shot, open weights | 0.377 | 0.620 | 0.320 | 0.203 | – | open weights |
|
| 108 |
| – | Prior (ignores the input) | reference | 0.470 | 0.347 | 0.189 | 0.088 | – | – |
|
| 109 |
| – | Uniform (same probability on every option) | reference | 0.308 | 0.444 | 0.238 | 0.169 | – | – |
|
| 110 |
|
|
|
|
| 117 |
‡ Reported by the submitter on their own hardware (a different GPU and a different unit for each), not comparable with the other rows.
|
| 118 |
§ Self-reported by the submitter in the linked discussion and not re-run by us. Metric definitions follow the submitter's report; ECE in particular is computed differently by different submitters.
|
| 119 |
◆ Scored by a third party ([SaylorTwift](https://huggingface.co/SaylorTwift), discussion #37) on Hugging Face Jobs, 2026-10-16 to 10-19, with one public scorer for every row and not re-run by us. Raw predictions and scripts are in the [results bucket](https://huggingface.co/buckets/SaylorTwift/typed-decisions-evals). Its accuracy counts a tie in the gold as correct and its ECE uses 10 bins, so small differences from the other rows can come from the scorer. Latency was measured on mixed Jobs hardware and is left out. d3-turbo was served through its own `system_one()` runtime because vLLM does not load it yet.
|
| 120 |
+
♠ One question per request, not one request per case. GEV-26B and JEV-27B were scored through the model's native `POST /v1/decide`. The three rows served by [Decision Models](https://decisionmodels.io) were scored by us on 2026-10-22 through its hosted `POST /v1/systemone`, which takes one question per call, with a key Decision Models provided ([#38](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/38)); their latency is p50 per case, i.e. five requests in a row. Each model is named after the open weights Decision Models says it serves.
|
| 121 |
|
| 122 |
### Fitted or fine-tuned on `train`
|
| 123 |
|
|
|
|
| 162 |
|
| 163 |
◇ Clean re-score by the submitter after removing 500 training pairs derived from model predictions on `test` rows ([discussion #25](https://huggingface.co/datasets/LocalLLaMA/typed-decisions/discussions/25)). The originally submitted 0.906 was measured on a split the model had partly seen, so it is not listed; KL, Brier and ECE were only reported for that run. Do not train on `test` rows, `test` labels, or anything derived from predictions on them.
|
| 164 |
|
| 165 |
+
**Quyet-1.0-Large, served by Decision Models, has the highest accuracy among zero-shot models (0.791, scored one question per request). [meraGPT Decider 1](https://meragpt.com/models/state-decider-1?utm_source=huggingface&utm_medium=dataset_card&utm_campaign=typed-decisions) is second on accuracy (0.768) and has the lowest KL (0.096) and Brier (0.052) of any zero-shot model, answering a whole case in one request.** It leads Jev on every question type (`noul` 0.840 vs
|
|
|
|
| 166 |
0.775, `choice` 0.733 vs 0.720, `score` 0.739 vs 0.696), its distributions sit far
|
| 167 |
closer to the gold (KL 0.096 vs 1.442), it is faster end to end, and it costs less
|
| 168 |
per token. It answers `POST /v1/systemone` at
|
assets/leaderboard.png
CHANGED
|
Git LFS Details
|
|
Git LFS Details
|