Skip to content

Benchmark

BABILong qa1 16k

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

Models and systems measured
13
Scores recorded
14
Sources
1
IQ weight
No weight published

Score distribution

66.6%
Mean
63%
Median
21.8%
Std dev
32%
10th percentile
100%
90th percentile
31%
Min
31%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-tuneqa1, 16k· BABILong v0 data docs100%
~ Mamba (130M) fine-tuneqa1, 16k· BABILong v0 data docs100%
~ RMT (137M) fine-tuneqa1, 16k· BABILong v0 data docs100%
ChatGLM3 6Bchatglm3-6b-128k, qa1, 16k, 128k· BABILong v0 data docs56%
GPT-4-0125-previewqa1, 16k· BABILong v0 data docs84%
Llama 2 7bLLaMA-2-7B-32K, qa1, 16k, 32k· BABILong v0 data docs32%
Llama3-ChatQA-1.5-8B + RAGqa1, 16k· BABILong v0 data docs60%
Mistral 7B Instruct v0.2Mistral-7b-Instruct-v0.2, qa1, 16k· BABILong v0 data docs57%
Mixtral 8x22B Instruct v0.1Mixtral-8x22B-Instruct-v0.1, qa1, 16k· BABILong v0 data docs63%
Mixtral 8x7B Instruct v0.1Mixtral-8x7B-Instruct-v0.1, qa1, 16k· BABILong v0 data docs53%
Phi-3 Medium 128K InstructPhi-3-medium-128k-instruct, qa1, 16k· BABILong v0 data docs72%
Phi-3 Mini 128K InstructPhi-3-mini-128k-instruct, qa1, 16k· BABILong v0 data docs60%
Yi-34B01-ai/Yi-34B-200k, qa1, 16k, 200k· BABILong v0 data docs65%
Yi-34B01-ai/Yi-34B, qa1, 16k· BABILong v0 data docs31%

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