metadata
datasets:
- open-web-math/open-web-math
QuantFactory/quietstar-8-ahead-GGUF
This is quantized version of ezelikman/quietstar-8-ahead created using llama.cpp
Original Model Card
Mistral-7b with continued pretraining using Quiet-STaR (https://arxiv.org/abs/2403.09629) for generating 8 thought tokens before each output token.