Hey, it's probably delusional.
#2
by Eilian - opened
Is it still possible to add version 2.7B to the model by now? It seems to have more behavior than Erebus can recognize.
best bet is lit3b-4k
same dataset (i think) but with 4k tokens of context
best bet is lit3b-4k
same dataset (i think) but with 4k tokens of context
But I'm afraid my video memory size prevents me from using it, and its 4bit version doesn't seem to exist anymore.
best bet is lit3b-4k
same dataset (i think) but with 4k tokens of contextBut I'm afraid my video memory size prevents me from using it, and its 4bit version doesn't seem to exist anymore.
you could join the server, there is some progress for quantizing it