CodeLlama-34B-Python-fp16: benchmark results
TheBloke's Transformers-format FP16 conversion of Meta's CodeLlama-34B-Python, converted from Meta's original weights. The Instruct and Python variants remain separate releases. Provider: TheBloke. Access: Open.
Aggregate Rating 1197 ± 17, rank #9633 of 10916 rated models, from 68 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Chemistry | 42 | Accuracy (%) | 80.4 |
| MMLU-by-task - College Physics | 28.43 | Accuracy (%) | 67.4 |
| MMLU-by-task - Formal Logic | 32.54 | Accuracy (%) | 59.5 |
| MMLU-by-task - College Mathematics | 32 | Accuracy (%) | 53.7 |
| MMLU-by-task - Moral Scenarios | 26.37 | Accuracy (%) | 53.7 |
| MMLU-by-task - Abstract Algebra | 30 | Accuracy (%) | 52.2 |
| MMLU-by-task - TruthfulQA MC1 | 27.78 | Accuracy (%) | 47.8 |
| MMLU-by-task - TruthfulQA MC2 | 43.57 | Accuracy (%) | 43.9 |
| MMLU-by-task - World Religions | 55.56 | Accuracy (%) | 42.1 |
| MMLU-by-task - High School Microeconomics | 36.13 | Accuracy (%) | 41.2 |
| Open LLM Leaderboard v1 - GSM8K | 20.02 | Accuracy (%) (5-shot) | 41 |
| MMLU-by-task - Management | 46.6 | Accuracy (%) | 40.9 |
Interactive version: theaggregate.ai/model?slug=codellama-34b-python-fp16 · How It Works · Data refreshed daily, snapshot 2026-11-08.