Python env for LLM Starter Kit

[Pierre_BEJIAN]

Hi,
I am working on LLM Starter Kit (compatible with DSS v12) and my question is simple: which Python envs do we need do run the LLM Starter Kit ?

Regards,

Pierre BEJIAN

---

Operating system used: Windows

[Emma]

Hey @Pierre_BEJIAN , 

All of the requirements can be found in the Project Wiki under "3.1 Technical requirements" section.  

But I will post the current list here: 

• requires to create a Python 3.10 (or ≥ 3.8.1 if LMQL is not used) code environment named py_310_sample_llm with the following packages:

  openai==0.27.8
  tiktoken==0.4.0
  anthropic==0.3.8
  cohere==4.19.2
  ai21==1.2.4
  lmql==0.0.6.5
  guidance==0.0.54
  langchain==0.0.234
  torch==2.0.1
  transformers==4.31.0
  sentence-transformers==2.2.2
  accelerate==0.21.0
  evaluate==0.4.0
  bert-score==0.3.13
  faiss-cpu==1.7.4
  pypdf==3.15.0
  unstructured[local-inference]==0.9.1
  tabulate==0.9.0
  dash==2.11.1
  dash-bootstrap-components==1.4.2