RedPajama-INCITE-Instruct-3B-v1: benchmark results
Together's RedPajama-INCITE-Instruct-3B-v1, a 2.8B model fine-tuned from RedPajama-INCITE-Base-3B-v1 for few-shot applications on GPT-JT data, developed with Ontocord.ai, ETH DS3Lab, Stanford CRFM and LAION. Provider: Together. Released 2023-05-05. Access: Open.
Aggregate Rating 1159 ± 13, rank #1601 of 1640 rated models, from 97 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Classic - Dyck | 70.8 | Exact Match (%) | 86.8 |
| HELM Classic - bAbI | 52.21 | Exact Match (%) | 72.5 |
| HELM Classic - RAFT | 66.14 | Exact Match (%) | 67.4 |
| HELM Classic - NaturalQuestions Open Book | 63.71 | F1 (%) | 64.6 |
| HELM Classic - Entity Matching | 82.79 | Exact Match (%) | 59.1 |
| MMLU-by-task - Global Facts | 34 | Accuracy (%) | 58.4 |
| HELM Classic - CivilComments | 54.95 | Exact Match (%) | 57.6 |
| HELM Classic - Entity Data Imputation | 75.24 | Exact Match (%) | 51.5 |
| HELM Classic - NarrativeQA | 63.76 | F1 (%) | 45.4 |
| MMLU-by-task - College Chemistry | 28 | Accuracy (%) | 35 |
| Open LLM Leaderboard - MuSR | 38.86 | Score | 34.5 |
| MMLU-by-task - High School Mathematics | 25.93 | Accuracy (%) | 34.1 |
Interactive version: theaggregate.ai/model?slug=redpajama-incite-instruct-3b-v1 · How It Works · Data refreshed daily, snapshot 2026-10-13.