lm_head missing?
#2
by Alvant - opened
Hello
It seems that there is no lm_head.weight in saved model parameters. Is it OK?) π
What is the proper way to load the model? Is it necessary to get this lm_head from somewhere else or something?..
Sorry! my fault. I didn't notice that the architecture is MixtralModel and not MixtralModelForCausalLM. Everything is OK then)
Alvant changed discussion status to closed