model_id stringclasses 3
values | lab stringclasses 2
values | release_date timestamp[s]date 2026-09-30 00:00:00 2026-10-05 00:00:00 | source stringclasses 3
values | benchmarks listlengths 23 33 |
|---|---|---|---|---|
LiquidAI/d1-3B | Liquid AI | 2026-10-05T00:00:00 | https://huggingface.co/LiquidAI/d1-3B | [
{
"name": "Decision Index 0.2.1",
"category": "aggregate"
},
{
"name": "DecisionBench",
"category": "decision"
},
{
"name": "Fast Decisions",
"category": "decision"
},
{
"name": "SQuAD 2.0",
"category": "qa_reading"
},
{
"name": "Civil Comments",
"category": "... |
autotrust/JEV-27B-VL | AutoTrust AI | 2026-09-30T00:00:00 | https://huggingface.co/autotrust/JEV-27B-VL | [
{
"name": "Jev Decision Index 0.3 — Vision board",
"category": "aggregate"
},
{
"name": "CVBench",
"category": "vision"
},
{
"name": "BLINK",
"category": "vision"
},
{
"name": "RealWorldQA",
"category": "vision"
},
{
"name": "CharXiv",
"category": "vision"
}... |
autotrust/GEV-26B-Decide | AutoTrust AI | 2026-10-02T00:00:00 | https://huggingface.co/autotrust/GEV-26B-Decide | [
{
"name": "Decision Index 0.2.1",
"category": "aggregate"
},
{
"name": "GPQA-Diamond",
"category": "reasoning"
},
{
"name": "CRUXEval",
"category": "reasoning"
},
{
"name": "CLadder",
"category": "reasoning"
},
{
"name": "GSM8K",
"category": "math"
},
{
... |
Decision Model Benchmark Usage
Which evaluation benchmarks labs use to evaluate their decision models (single-pass,
zero-output-token classifiers/raters — routing, triage, moderation, reranking, LLM-judge,
visual inspection — not chat LLMs). Sister dataset to
SaylorTwift/llm-benchmark-usage,
which covers LLM/agent evaluation suites; benchmark names are canonicalized against
that dataset so overlapping benchmarks (e.g. AI2D, MMBench, BoolQ) share one spelling
across both.
Hand-built from papers, technical reports, system cards, model cards, and blog posts.
from datasets import load_dataset
models = load_dataset("SaylorTwift/decision-model-benchmark-usage", "models")["models"]
models
One row per model, fully self-contained.
| column | type | description |
|---|---|---|
model_id |
string | Hugging Face repo id for open-weight models (e.g. LiquidAI/d1-3B), or a plain slug for closed/API models |
lab |
string | Organization/lab that released the model |
release_date |
timestamp | Release date. HF repo creation date for open-weight models (a proxy, not always the exact announcement date); hand-researched announcement date for closed models |
source |
string | Link to the paper/report/card/blog this model's evaluation suite was extracted from |
benchmarks |
list | The evaluation suite: {name, category} objects. Names are canonicalized (e.g. always CVBench, never CV-Bench); genuinely distinct variants are kept separate (SQuAD vs SQuAD 2.0). Categories are per-source free-text groupings, not a controlled vocabulary |
Richer per-source metadata (paper type, title, notes, confidence flags) lives in a local
sources table and is not published as of this schema version; it may move to a separate
Hub dataset later.
- Downloads last month
- 34