Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/aaltorse/llm-cli

Simple OpenAI APi compatible terminal interface in Python

https://github.com/aaltorse/llm-cli

Science Score: 13.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ○
    .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (12.1%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Simple OpenAI APi compatible terminal interface in Python

Basic Info
  • Host: GitHub
  • Owner: AaltoRSE
  • Language: Python
  • Default Branch: master
  • Size: 29.3 KB
Statistics
  • Stars: 0
  • Watchers: 2
  • Forks: 1
  • Open Issues: 0
  • Releases: 0
Created over 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme

README.md

llm.py

This serves as a command line shell to an OpenAI-like completion service. The purpose is to query the API from the command line.

Features include: saving per-thread history to topics (and re-loading history from those states), history files are YAML so easy to edit and re-test), use custom APIs (not just OpenAI), adjust all the main parameters (model, temperature, system prompt), parameters per-thread persisted to the thread files, global config file.

Installation

llm.py is self-contained other than the fairly standard dependencies requests and pyyaml.

Usage

python llm.py [-t thread_history.yaml]

Options are loaded in this order, last one takes precedence. Whatever is loaded gets saved back to the history file: * Build-in (see params dict) * File specified with --config, -c * Loaded from history with -thread, -t * Specified on command line.

Options:

  • --thread, -t: Save state to this YAML file and read from it. It will also save some of the main parameters below. This can be used to resume (and edit) chats and reply them.
  • --system: The system prompt.
  • --temperature, --model: as expected
  • --max-tokens: Max model input tokens (save some for the output)
  • --config: A config file with defaults.
  • --list-models, -l: Don't do anything but list the models.

More options are explained by using --help.

The config file syntax is a YAML file containing a mapping from values above to their values. (but max_tokens instead of max-tokens).

Status

This was mainly made for one person's work and might not be that useful to others (and partly made as an experiment for learning). Still the threads part seems unique. I'd welcome contributions or feature requests.

See also

  • https://github.com/kardolus/chatgpt-cli - the first one I used
  • https://github.com/simonw/llm - next thing I was recommended to use. But didn't quite do what I want, at least I didn't see it when I looked. This might be what you are looking for instead.

Owner

  • Name: AaltoRSE
  • Login: AaltoRSE
  • Kind: organization

GitHub Events

Total
  • Push event: 1
Last Year
  • Push event: 1