Skip to content

Benchmark

BABILong qa5 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

74.9%
Mean
75%
Median
14.8%
Std dev
55%
10th percentile
98.1%
90th percentile
51%
Min
51%99%
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-tuneqa5, 16k· BABILong v0 data docs98.1%
~ Mamba (130M) fine-tuneqa5, 16k· BABILong v0 data docs99%
~ RMT (137M) fine-tuneqa5, 16k· BABILong v0 data docs94%
ChatGLM3 6Bchatglm3-6b-128k, qa5, 16k, 128k· BABILong v0 data docs61%
GPT-4-0125-previewqa5, 16k· BABILong v0 data docs85%
Llama 2 7bLLaMA-2-7B-32K, qa5, 16k, 32k· BABILong v0 data docs51%
Llama3-ChatQA-1.5-8B + RAGqa5, 16k· BABILong v0 data docs77%
Mistral 7B Instruct v0.2Mistral-7b-Instruct-v0.2, qa5, 16k· BABILong v0 data docs67%
Mixtral 8x22B Instruct v0.1Mixtral-8x22B-Instruct-v0.1, qa5, 16k· BABILong v0 data docs64%
Mixtral 8x7B Instruct v0.1Mixtral-8x7B-Instruct-v0.1, qa5, 16k· BABILong v0 data docs67%
Phi-3 Medium 128K InstructPhi-3-medium-128k-instruct, qa5, 16k· BABILong v0 data docs82%
Phi-3 Mini 128K InstructPhi-3-mini-128k-instruct, qa5, 16k· BABILong v0 data docs73%
Yi-34B01-ai/Yi-34B-200k, qa5, 16k, 200k· BABILong v0 data docs75%
Yi-34B01-ai/Yi-34B, qa5, 16k· BABILong v0 data docs55%

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