Dequantized Base Models
Collection
Dequantized (fp16) versions of nf4 quantized base models for faster inference.
•
27 items
•
Updated
•
1
This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.