https://huggingface.co/recursal/QRWKV6-32B-Instruct-Preview-v0.1
#552
by
nicoboss
- opened
This is the first model using the QRWKV6 architecture
QRWKV6 is a linear and not a transformer based archidecture.
It seams to perform better than Qwen2.5-32B from which it is based in most beanchmarks.
Make sure to use at llatest llama.cpp or it will not work as support for QRWKV6 got merged 12 hours ago
will queue soon :)
mradermacher
changed discussion status to
closed