NEBULA-XB-v1.0_SFT_2_epoch: benchmark results
A second-stage depth-upscaling release by TeeZee with 108 layers, derived through the SlimOrca-fine-tuned GALAXY v03 model. This SFT variant further fine-tunes NEBULA-XB-v1.0 on 30,000 Merge_Glue entries. Provider: TeeZee. Access: Open.
Aggregate Rating 1422 ± 36, rank #5056 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 65.41 | Accuracy (%) (5-shot) | 84.3 |
| Open LLM Leaderboard v1 - WinoGrande | 82.24 | Accuracy (%) (5-shot) | 83 |
| Open LLM Leaderboard v1 - HellaSwag | 85.07 | Normalized accuracy (%) (10-shot) | 71.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 63.05 | Normalized accuracy (%) (25-shot) | 61 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 52.06 | MC2 (%) (0-shot) | 54.4 |
| Open LLM Leaderboard v1 - GSM8K | 0.3 | Accuracy (%) (5-shot) | 11.4 |
Interactive version: theaggregate.ai/model?slug=nebula-xb-v1-0-sft-2-epoch · How It Works · Data refreshed daily, snapshot 2026-11-08.