This is a very low loss quantization ( should have >99% perplexity) of what was for a while the #1 model on the openllm leaderboard.
https://huggingface.co/dnhkng/RYS-XLarge
This model was a franken of qwen2 72b , and is very very likely to get surpased by qwen2.5 finetunes, so I'm more posting it for reference, but it's a great model nonetheless.
- Downloads last month
- 15
Model tree for nisten/rys-78b-55GB-gguf
Base model
dnhkng/RYS-XLarge