SimpleChat:DU:Avoid setting frequence/Presence penalty

Some models like llama3 found to try to be over intelligent by
repeating garbage still, but by tweaking the garbage a bit so that
it is not exactly same. So avoid setting these penalties and let
the model's default behaviour work out, as is.

Also the simple minded histogram based garbage trimming from end,
works to an extent, when the garbage is more predictable and
repeatative.
This commit is contained in:
HanishKVC 2024-05-26 22:09:32 +05:30
parent a41f701159
commit ed345abac8

View file

@ -540,9 +540,9 @@ class Me {
this.chatRequestOptions = { this.chatRequestOptions = {
"temperature": 0.7, "temperature": 0.7,
"max_tokens": 1024, "max_tokens": 1024,
"frequency_penalty": 1.2, "n_predict": 1024,
"presence_penalty": 1.2, //"frequency_penalty": 1.2,
"n_predict": 1024 //"presence_penalty": 1.2,
}; };
} }