zephyr-220m-sft-full: benchmark results
BEE-spoke-data release: A supervised fine-tune of smol_llama-220M-openhermes on Ultrachat_200k, distributed with a MistralForCausalLM configuration; it is distinct from the later preference-trained release. Provider: BEE-spoke-data. Access: Open.
Aggregate Rating 1108 ± 42, rank #10626 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 43.23 | MC2 (%) (0-shot) | 23 |
| Open LLM Leaderboard v1 - MMLU | 26.45 | Accuracy (%) (5-shot) | 12.6 |
| Open LLM Leaderboard v1 - GSM8K | 0.38 | Accuracy (%) (5-shot) | 11.9 |
| Open LLM Leaderboard v1 - WinoGrande | 51.62 | Accuracy (%) (5-shot) | 5.7 |
| Open LLM Leaderboard v1 - HellaSwag | 29.03 | Normalized accuracy (%) (10-shot) | 5 |
| Open LLM Leaderboard v1 - ARC Challenge | 25.26 | Normalized accuracy (%) (25-shot) | 4.6 |
Interactive version: theaggregate.ai/model?slug=zephyr-220m-sft-full · How It Works · Data refreshed daily, snapshot 2026-11-08.