feat: add dolly/redpajama/bloomz models support

[mudler]

This adds experimental supports for dolly models and red pajamas, however I didn't had still have success in running any of the models I've found:

main: seed = 1683660204
dollyv2_model_load: loading model from '/home/mudler/_git/LocalAI/models/ggml-dolly-q5_0.bin' - please wait ...
dollyv2_model_load: n_vocab = 50280
dollyv2_model_load: n_ctx   = 2048
dollyv2_model_load: n_embd  = 4096
dollyv2_model_load: n_head  = 32
dollyv2_model_load: n_layer = 32
dollyv2_model_load: n_rot   = 32
dollyv2_model_load: ftype   = 8
dollyv2_model_load: ggml ctx size = 8596.22 MB
dollyv2_model_load: memory_size =  1024.00 MB, n_mem = 65536
dollyv2_model_load: unknown tensor 'gpt_neox.embed_in.weight' in model file
main: failed to load model from '/home/mudler/_git/LocalAI/models/ggml-dolly-q5_0.bin'

However, I get the same error while running the main.cpp in ggml too.

Bloomz (https://github.com/NouamaneTazi/bloomz.cpp) seems to work, but no good model found so far.

[mudler]

Next on my list is to put gpt4all-j on par with the bindings