Bielik-11B-v2.0-Instruct: benchmark results
SpeakLeash and ACK Cyfronet AGH's Polish 11B instruct tune of Bielik-11B-v2, a depth-upscaled Mistral 7B v0.2 trained on curated Polish corpora (August 2024). Provider: SpeakLeash. Released 2024-08-26. Access: Open.
Aggregate Rating 1513 ± 16, rank #742 of 1656 rated models, from 74 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open PL LLM - PolEmo2-IN (multiple choice, 5-shot) | 88.78 | Accuracy (%) | 99.4 |
| Open PL LLM - PSC (multiple choice, 5-shot) | 94.45 | Binary F1 (%) | 98.7 |
| Open Portuguese LLM - FaQuAD NLI | 83.75 | Macro F1 (%) | 97.1 |
| Open PL LLM - PolQA Open Book (generative, 5-shot) | 91.92 | Levenshtein Similarity (%) | 96.7 |
| Open PL LLM - RAG | 75.5 | Average RAG Score (%) | 96.2 |
| Open PL LLM - Multiple Choice | 63.31 | Average Multiple-Choice Score (%) | 95.1 |
| Open PL LLM - PolEmo2-OUT (multiple choice, 5-shot) | 79.15 | Accuracy (%) | 92.2 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 79.5 | Accuracy (%) | 90.6 |
| Open PL LLM Leaderboard | 65.12 | Average Score (%) | 90.2 |
| Open PL LLM - PolQA Reranking (multiple choice, 5-shot) | 82.45 | Accuracy (%) | 89.9 |
| Open PL LLM - PolQA Closed Book (generative, 5-shot) | 82.45 | Levenshtein Similarity (%) | 89 |
| Open PL LLM - DYK (multiple choice, 5-shot) | 68.61 | Binary F1 (%) | 88.6 |
Interactive version: theaggregate.ai/model?slug=bielik-11b-v2-0-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.