Doesn't work well at all.
#1
by
Delcos
- opened
I've raised the concern on the ui page and in a discord channel but this model preforms FAR worse when in GGUF, even using the same seeds to the point of writing sentences and then stopping midway through and just adding a . I've made sure it wasn't a download issue, so I'm not sure what would be causing the issue. I used the q6 versions for all of the tests on multiple models, including the GGLM version of this one.
It's been found to be an issue with the version of llama.cpp. oops
Delcos
changed discussion status to
closed