Skip to content

Benchmark

BABILong qa2 128k

Long context benchmarks measure how well models use information in long inputs. Scores are percentages. Higher is better.

Models and systems measured
8
Scores recorded
8
Sources
1
IQ weight
No weight published

Score distribution

34.4%
Mean
25%
Median
36.1%
Std dev
5%
10th percentile
87%
90th percentile
2%
Min
2%100%
The band covers the middle 80% of scores. The line marks the median.

Results

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
~ ARMT (137M) fine-tuneqa2, 128k· BABILong v0 data docs100%
~ Mamba (130M) fine-tuneqa2, 128k· BABILong v0 data docs87%
~ RMT (137M) fine-tuneqa2, 128k· BABILong v0 data docs39%
ChatGLM3 6Bchatglm3-6b-128k, qa2, 128k· BABILong v0 data docs5%
GPT-4-0125-previewqa2, 128k· BABILong v0 data docs25%
Llama3-ChatQA-1.5-8B + RAGqa2, 128k· BABILong v0 data docs6%
Phi-3 Medium 128K InstructPhi-3-medium-128k-instruct, qa2, 128k· BABILong v0 data docs11%
Phi-3 Mini 128K InstructPhi-3-mini-128k-instruct, qa2, 128k· BABILong v0 data docs2%

Data credits: RMT-team, Kuratov et al. Official v0 numeric CSV; results licence not separately stated.