saiga2_7b_gguf / README.md
IlyaGusev's picture
Update README.md
464aadc
metadata
datasets:
  - IlyaGusev/ru_turbo_alpaca
  - IlyaGusev/ru_turbo_saiga
  - IlyaGusev/ru_sharegpt_cleaned
  - IlyaGusev/oasst1_ru_main_branch
  - IlyaGusev/ru_turbo_alpaca_evol_instruct
  - lksy/ru_instruct_gpt4
language:
  - ru
inference: false
pipeline_tag: conversational
license: llama2

Llama.cpp compatible versions of an original 7B model.

Download one of the versions, for example model-q4_K.gguf.

wget https://huggingface.co/IlyaGusev/saiga2_7b_gguf/resolve/main/model-q4_K.gguf

Download interact_llamacpp.py

wget https://raw.githubusercontent.com/IlyaGusev/rulm/master/self_instruct/src/interact_llamacpp.py

How to run:

pip install llama-cpp-python fire

python3 interact_llamacpp.py model-q4_K.gguf

System requirements:

  • 10GB RAM for q8_0 and less for smaller quantizations