Datasets:
The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 236, in _generate_tables
pa_table = paj.read_json(
^^^^^^^^^^^^^^
File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: JSON parse error: Column(/questions/[]/ground_truth) changed from number to string in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 93, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 250, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/utils/json.py", line 90, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 65, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
World Awareness Benchmark
Does your AI agent know what's happening in the world right now?
This benchmark tests whether agents can accurately answer factual questions about current events — recession probability, geopolitical risk, oil prices, election odds. Ground truth comes from prediction markets, where participants risk real money on outcomes.
Why This Benchmark Exists
LLMs have a knowledge cutoff. When asked "what's the probability of a US recession?", they either:
- Hallucinate a number
- Hedge with "I don't have access to real-time data"
- Give a stale figure from training data
There was no standard way to measure this gap. Now there is.
Format
{
"id": "geo-001",
"question": "What is the SF Geopolitical Risk Index score (0-100)?",
"category": "Geopolitical",
"ground_truth": 62,
"unit": "score_0_100",
"source": "SF Index",
"tolerance": 5,
"date": "2026-04-02"
}
Categories
| Category | Questions | What it tests |
|---|---|---|
| Geopolitical | 10 | War probabilities, nuclear risk, regional conflicts |
| Economy | 10 | Recession, Fed rates, market indices, treasury yields |
| Energy | 7 | Oil prices, energy contracts, supply disruption |
| Elections | 5 | Presidential, senate, political outcomes |
| Markets | 12 | Crypto, tech, gold, mispriced edges |
Scoring
| Result | Points |
|---|---|
| Exact match (within tolerance) | 2 |
| Correct direction/range | 1 |
| Wrong, hallucinated, or refused | 0 |
Max score: questions × 2
How to Use
import json
import requests
# Load benchmark
bench = json.load(open("benchmark_2026-04-02.json"))
# Test your agent
score = 0
for q in bench["questions"]:
answer = your_agent(q["question"]) # your agent's response
# Parse numeric answer and compare to ground_truth with tolerance
# ...
print(f"World Awareness Score: {score}/{bench['scoring']['max_score']}")
With SimpleFunctions (baseline)
An agent using SimpleFunctions world state should score near-perfect:
from simplefunctions import world
state = world()
# Inject into system prompt, then answer questions
# Expected score: 80-90% of max
Without any world context (baseline)
A vanilla LLM with no real-time data:
# Expected score: 0-10% of max (hallucinations occasionally correct by chance)
Monthly Updates
This benchmark is regenerated monthly from live prediction market data. Each file is dated (benchmark_YYYY-MM-DD.json). Ground truth changes as the world changes — that's the point.
Data Source
SimpleFunctions — 9,706 prediction market contracts from Kalshi (CFTC-regulated) and Polymarket. Calibrated by real money.
Citation
@dataset{simplefunctions_world_awareness_2026,
title={World Awareness Benchmark: Testing AI Agent Knowledge of Current Events},
author={SimpleFunctions},
year={2026},
url={https://huggingface.co/datasets/SimpleFunctions/world-awareness-bench},
note={Ground truth from 9,706 prediction markets. Updated monthly.}
}
License
MIT
- Downloads last month
- 29