Eval request
I tested the first model. I suppose I'll wait for the rest to get quanted.
@DontPlanToEnd
Hello, I have this question, is there any limit to how many models in what period you can send, or not?
Not really, but if all the models being requested are made by the person requesting their testing, I'd prefer to test at most like 10-15 per week. I don't want like half the models on the leaderboard to be 9b merges made by the same person, y'know.
I'll close this for now since the rest haven't been quanted yet. Hopefully when I move over from llama-cpp-python to aphrodite I won't need gguf quants in order to test models, I'll just use something like fp6.
@DontPlanToEnd
As I understand it you need to translate models into GGUF yourself?
As I understand it you need to translate models into GGUF yourself?
I've never made a gguf before, so I use ones others make. I wouldn't want to mess something up.
@DontPlanToEnd
There is a fairly simple space for this, ggml-org/gguf-my-repo, you can create GGUFs for many models there without much knowledge.