Benchmark
BABILong avg(qa1-5) 32k
Long context benchmarks measure how well models use information in long inputs. Scores are percentages. Higher is better.
All long context benchmarksScores from BABILong v0 data docs.
- Models and systems measured
- 15
- Scores recorded
- 16
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 49.5%
- Mean
- 48%
- Median
- 25.6%
- Std dev
- 36%
- 10th percentile
- 98%
- 90th percentile
- 3%
- Min
Results
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| ~ ARMT (137M) fine-tuneavg, 32k· BABILong v0 data docs | 98% |
| ~ Mamba (130M) fine-tuneavg, 32k· BABILong v0 data docs | 98.1% |
| ~ RMT (137M) fine-tuneavg, 32k· BABILong v0 data docs | 77.9% |
| ChatGLM3 6Bchatglm3-6b-128k, avg, 32k, 128k· BABILong v0 data docs | 36% |
| GPT-4-0125-previewavg, 32k· BABILong v0 data docs | 53% |
| Llama 2 7bLLaMA-2-7B-32K, avg, 32k· BABILong v0 data docs | 3% |
| Llama 3.1 70B InstructMeta-Llama-3.1-70B-Instruct, avg, 32k· BABILong v0 data docs | 59% |
| Llama 3.1 8B InstructMeta-Llama-3.1-8B-Instruct, avg, 32k· BABILong v0 data docs | 56% |
| Llama3-ChatQA-1.5-8B + RAGavg, 32k· BABILong v0 data docs | 44% |
| Mistral 7B Instruct v0.2Mistral-7b-Instruct-v0.2, avg, 32k· BABILong v0 data docs | 37% |
| Mixtral 8x22B Instruct v0.1Mixtral-8x22B-Instruct-v0.1, avg, 32k· BABILong v0 data docs | 43% |
| Mixtral 8x7B Instruct v0.1Mixtral-8x7B-Instruct-v0.1, avg, 32k· BABILong v0 data docs | 40% |
| Phi-3 Medium 128K InstructPhi-3-medium-128k-instruct, avg, 32k· BABILong v0 data docs | 53% |
| Phi-3 Mini 128K InstructPhi-3-mini-128k-instruct, avg, 32k· BABILong v0 data docs | 42% |
| Yi-34B01-ai/Yi-34B-200k, avg, 32k, 200k· BABILong v0 data docs | 48% |
| Yi-34B01-ai/Yi-34B, avg, 32k· BABILong v0 data docs | 4% |
Data credits: RMT-team, Kuratov et al. Official v0 numeric CSV; results licence not separately stated.