Skip to main content

Llama 3.3 70B, raw

GlobalOpinionQA · Raw LLM · default settings · 100 questions

raw--llama-3.3-70b-instruct--tdefault--tplcurrent--f2748030

Compare with another GlobalOpinionQA config →

Metadata

Raw LLM · 2 replicates · 100 questions
Method
Raw LLM
Base provider
meta
Model
llama-3.3-70b-instruct
Dataset
GlobalOpinionQA
Temperature
default
Reasoning effort
default
Template
current
Samples / question
15
Questions
100
First replicate
2026-04-12 01:48 UTC
Last replicate
2026-07-15 17:33 UTC

Aggregated metrics

Mean across replicates, ± sample standard deviation (shown only with 2+ replicates). SynthBench Index: 0 = random answers, 100 = real people. SPS is the raw 0 to 1 score it is computed from. Index = 100 × (SPS − SPS of random answers) ÷ (human ceiling − SPS of random answers), on this config's dataset. The leaderboard averages it over the core datasets.

Index
0 = random answers, 100 = real people
-4
± 2.3
SPS
SPS, 0 to 1
0.766
± 0.005
Distribution gap
mean JSD, lower is better
0.361
± 0.005
Rank agreement
Kendall τ, −1 to 1
0.358
± 0.020

Variance summary

Replicates
2
Std dev
2.3 Index points (SPS 0.005)
Range
Index -5 to -2 (SPS 0.762–0.770)
Coefficient of variation (SPS)
0.0067

Per-topic breakdown

Index per topic (mean across replicates; 0 = random answers, 100 = real people). All seven topics are shown; topics this config's questions don't cover are marked "no questions".

Show data table

Loading data…

Replicates (2)

Run Timestamp Index SPS JSD Kendall τ N
globalopinionqa_openrouter_meta-llama_llama-3.3-70b-instruct_20260412_014837 2026-04-12 01:48 UTC -5 0.762 0.365 0.344 100
globalopinionqa_openrouter_meta-llama_llama-3.3-70b-instruct_20260715_173346 2026-07-15 17:33 UTC -2 0.770 0.357 0.372 100