'We have EnSEmblEd toP RAnKeRs tO mAKE oUr mODeL.'
'This MoDeL hAS BeEN trAINED oN orCa-StyLe DatAsEtS.'
'trAINed' KEK
'We hAVE AchIEvEd TOp RAnKeR In thE leAdErBOArDs.'
Lmao its all bs, you're all running merge scripts like we do, literally trying to game the leaderboards huh? Pathetic lol. Literally 0 card info, copy pasted from llama2 base models, with no other goals than going for leaderboards.
Atleast merge with a goal, like come on. I'm going for RP, Undi's going for RP. Atleast make a believable goal.
Brainderp is exposing the frauds who lie about it lol. Atleast be honest if you're merging like us poor people, which I appreciate some models do.
for BrainDerp, I have 'ensembled' several random models in the leaderboard to create this model.
i can easily steal the top leaderboard spots but meh not worth the effort.
that's it. i didn't bother testing much. ymmv.
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_Sao10K__BrainDerp)
| Metric | Value |
|---|---|
| Avg. | 51.16 |
| ARC (25-shot) | 60.75 |
| HellaSwag (10-shot) | 82.1 |
| MMLU (5-shot) | 58.81 |
| TruthfulQA (0-shot) | 56.9 |
| Winogrande (5-shot) | 75.85 |
| GSM8K (5-shot) | 8.26 |
| DROP (3-shot) | 15.48 |
- Downloads last month
- 3,091