Y
Hacker News
new
|
ask
|
show
|
jobs
by
peakji
603 days ago
It is an LLM fine-tuned using a new type of dataset and RL reward. It's good at reasoning, but I would not recommend to replace Llama for general tasks.