Quantizations of https://huggingface.co/ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.1
Inference Clients/UIs
From original readme
ArliAI-RPMax-12B-v1.1 is a variant based on Mistral Nemo 12B Instruct 2407.
This is arguably the most successful RPMax model due to how Mistral is already very uncensored in the first place.
Training Details
- Sequence Length: 8192
- Training Duration: Approximately 2 days on 2x3090Ti
- Epochs: 1 epoch training for minimized repetition sickness
- QLORA: 64-rank 128-alpha, resulting in ~2% trainable weights
- Learning Rate: 0.00001
- Gradient accumulation: Very low 32 for better learning.
Suggested Prompt Format
Mistral Instruct Prompt Format
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 20.64 |
| IFEval (0-Shot) | 53.49 |
| BBH (3-Shot) | 24.81 |
| MATH Lvl 5 (4-Shot) | 9.21 |
| GPQA (0-shot) | 4.25 |
| MuSR (0-shot) | 5.56 |
| MMLU-PRO (5-shot) | 26.49 |
- Downloads last month
- 626
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model authors have turned it off explicitly.