lbarasc

How to create customised vocal models to use with train-text-from-scratch ?

1

I made some experiments with train-text-from-scratch, we must specify a vocab model, where to find it or how to create it ? I find the file ggml-vocab-llama.gguf on hugging face,...

Support website hosting

2

How to keep always the some .onion address when share website ? If i stop the app (when i shut down device) and restart it the onion address isn't always...

hi, i dowloaded some FLUX models but when i want to delete it from the user interface of stabilitymatrix, i cannot, nothing happens. i manualy remove models from the directory...

Results about others supported 1-bit models ?

2

Do you test the others 1-bit models from Hugginface ? bitnet_b1_58-large bitnet_b1_58-3B Llama3-8B-1.58-100B-tokens Falcon3 Family | 1B-10B What are the results in inference ? I personnaly use Falcon3-10B-Instruct-1.58bit.gguf and results...

local bitnet model don't have the same quality as online bitnet

3

It seems, the bitnet local model doesn't have the same quality as the online demo from microsoft site : https://bitnet-demo.azurewebsites.net/ What are the parameters for online demo model ? i...

Binaries releases for Windows ?

12

Hi, Can you release binaries for windows working on different types of CPU (avx,avx2 etc...) ? Thank you.