Does it actually work ?
I tried running it, but it is showing this error :
gguf_init_from_reader: tensor 't0' has invalid number of dimensions: 8 > 4
The model doesnt actually work
Its just made to display the largest possible amount of parameters in huggingface.
The error is:
Tensor 0 (and all other tensors) have 8 dimensions.
But the inference engine you use doesnt allow models to have more than 4 dimensions per tensor.
Q&a
why did You use 8 dimensions per tensor?
Answer: because its the maximum huggingface allows to not Mark the model as 'invalid' and therefore no longer display parameter count, but also large enough that the parameter count displayed is the hightest
Ah I see !
I tried running it, but it is showing this error :
gguf_init_from_reader: tensor 't0' has invalid number of dimensions: 8 > 4
Bro, 1e145 parameters is a lot, think about it: there is around, in the observable universe, 1e80 atoms, so it's 1e65 times less, it means that for each atoms in the universe, you would have to store actually 1000...65 zeros...000 parameters... Do you think it is possible to use all of them, and to do calculations?