Instructions to use mozilla-ai/rocket-3B-llamafile with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use mozilla-ai/rocket-3B-llamafile with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("mozilla-ai/rocket-3B-llamafile", dtype="auto") - Notebooks
- Google Colab
- Kaggle
How to prevent hallucinations
#2
by JLouisBiz - opened
Updated Settings:
Predictions: 2048
Temperature: 0.5
Penalize Repeat Sequence: 1.1
Consider N Tokens for Penalize: 256
Top-K Sampling: 50
Top-P Sampling: 0.9
Min-P Sampling: 1
I have tried updating those settings, but is not good. I wish to prevent hallucinations. All I need is text summary. But I get good begin of summary followed by large number of repetitive single words, or something.