Skip to content

Benchmark

BABILong qa4 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

59.9%
Mean
60%
Median
19.6%
Std dev
36%
10th percentile
100%
90th percentile
32%
Min
32%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-tuneqa4, 16k· BABILong v0 data docs100%
~ Mamba (130M) fine-tuneqa4, 16k· BABILong v0 data docs100%
~ RMT (137M) fine-tuneqa4, 16k· BABILong v0 data docs75%
ChatGLM3 6Bchatglm3-6b-128k, qa4, 16k, 128k· BABILong v0 data docs32%
GPT-4-0125-previewqa4, 16k· BABILong v0 data docs47%
Llama 2 7bLLaMA-2-7B-32K, qa4, 16k, 32k· BABILong v0 data docs36%
Llama3-ChatQA-1.5-8B + RAGqa4, 16k· BABILong v0 data docs60%
Mistral 7B Instruct v0.2Mistral-7b-Instruct-v0.2, qa4, 16k· BABILong v0 data docs58%
Mixtral 8x22B Instruct v0.1Mixtral-8x22B-Instruct-v0.1, qa4, 16k· BABILong v0 data docs60%
Mixtral 8x7B Instruct v0.1Mixtral-8x7B-Instruct-v0.1, qa4, 16k· BABILong v0 data docs61%
Phi-3 Medium 128K InstructPhi-3-medium-128k-instruct, qa4, 16k· BABILong v0 data docs61%
Phi-3 Mini 128K InstructPhi-3-mini-128k-instruct, qa4, 16k· BABILong v0 data docs50%
Yi-34B01-ai/Yi-34B-200k, qa4, 16k, 200k· BABILong v0 data docs56%
Yi-34B01-ai/Yi-34B, qa4, 16k· BABILong v0 data docs43%

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