TeeZee/GALAXY-XB-v.02
Experiment, can DUS be taken one or more steps further?
Technical notes:
- 10 layers removed from both models this time, 2 more than in original paper.
- base version of upstage/SOLAR-10.7B-v1.0 used for merge
- no finetuning done yet, this is just a merge, first step in DUS paper
- next step, if evaluation proves that its at least as 'smart' as base model, should be finetuning to 'recover' after merge
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 62.48 |
| AI2 Reasoning Challenge (25-Shot) | 60.67 |
| HellaSwag (10-Shot) | 83.27 |
| MMLU (5-Shot) | 64.99 |
| TruthfulQA (0-shot) | 43.60 |
| Winogrande (5-shot) | 80.27 |
| GSM8k (5-shot) | 42.08 |