llm-jp-13B-instruct-full-jaster-dolly-oasst-v1.0: benchmark results
LLM-jp v1 instruction-tuned release developed by NII's LLM-jp project, using GPT-2 architecture and a 2,048-token context window. The named instruction-tuning generation remains distinct from the pretrained checkpoint and other instruct generations. Provider: National Institute of Informatics. Access: Open.
Aggregate Rating 1133 ± 52, rank #10250 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| pfgen-bench - Completion Mode - Truthfulness | 0.79 | Truthfulness Score | 60.7 |
| pfgen-bench - Completion Mode - Score | 0.52 | pfgen Score (mean of three) | 59.7 |
| pfgen-bench - Completion Mode - Fluency | 0.61 | Fluency Score | 58.9 |
| pfgen-bench - Completion Mode - Helpfulness | 0.16 | Helpfulness Score | 58 |
| pfgen-bench - QA Mode - Truthfulness | 0.72 | Truthfulness Score | 43.9 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 45.19 | MC2 (%) (0-shot) | 30.1 |
| pfgen-bench - QA Mode - Helpfulness | 0.06 | Helpfulness Score | 21.8 |
| pfgen-bench - QA Mode - Score | 0.35 | pfgen Score (mean of three) | 19.4 |
| Open LLM Leaderboard v1 - HellaSwag | 44.78 | Normalized accuracy (%) (10-shot) | 8.9 |
| pfgen-bench - QA Mode - Fluency | 0.28 | Fluency Score | 7.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 26.88 | Normalized accuracy (%) (25-shot) | 6.1 |
| Open LLM Leaderboard v1 - GSM8K | 0 | Accuracy (%) (5-shot) | 4.5 |
Interactive version: theaggregate.ai/model?slug=llm-jp-13b-instruct-full-jaster-dolly-oasst-v1-0 · How It Works · Data refreshed daily, snapshot 2026-11-08.