|
--- |
|
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](https://huggingface.co/IlyaGusev/saiga2_7b_lora). |
|
|
|
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](https://raw.githubusercontent.com/IlyaGusev/rulm/master/self_instruct/src/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 |
|
|
