https://huggingface.co/Ateeqq/Text-Rewriter-Paraphraser
#1193
by
dylrob1
- opened
Still doing on going research into AI detection, going to compare pre trained paraphrasing against reinforcement trained models. This model has an Apache 2.0 license. Thank you!
I am still testing F32 across these models. Since this model is so small (.4b) I feel like F16 will actually negatively effect this model since it only has one specific use case and those edge cases will matter and that level of nuance is needed. From what I have found anything over 7b, F16 seems to not effect specific use cases as much. Where as the really small models seem to be more sensitive to precision level changes.
Never mind this one is really bad. Tested it out on CPU.
dylrob1
changed discussion status to
closed