File size: 906 Bytes
de1b021 26acd79 de1b021 32c502e 26acd79 32c502e 26acd79 32c502e |
1 2 3 4 5 6 7 8 9 10 11 12 |
---
license: gpl-3.0
language:
- en
---
These are some LoRAs you can use to make sure that LoRAs are loading the way you expect them to.
`AAAAAAA` and `BBBBBBB` can be used with with [Neko-Institute-of-Science/LLaMA-7B-4bit-128g](https://huggingface.co/Neko-Institute-of-Science/LLaMA-7B-4bit-128g)
`8bit_AAAAAAA` and `8bit_BBBBBBB` can be used with [decapoda-research/llama-7b-hf](https://huggingface.co/decapoda-research/llama-7b-hf)
These models do what is on the label `AAAAAAA` will almost always ouput a stream of the letter "A". `BBBBBBB` will do the same for "B".
I decided to make these because it was frustrating trying to figure out whether models were actually loading the way I expected them to. You could, instead, set your parameters and seed to the same thing and do a couple generation tests, with and without LoRA, and look for differences, but this is a lot easier to see at a glance. |