/graph_to_agent

the vision is as vivid as reality itself: a platform where people can share their knowledge as knoweldge graphs, journey through these universes of wisdom, and engage with modular agents that embody and discuss the clusters of the universes the user has selected.

Primary LanguageJupyter NotebookGNU General Public License v3.0GPL-3.0

graph_to_agent

For Vision-Statement see: Vision.md

Motivation

  1. Understand the motivation

Git Workflow Rules

  1. Branching Strategy:

    • Each developer should work on their own development branch (e.g. feature-... development branch).
    • The main branch should only be updated through pull requests.
    • Pull requests to the main branch require a review before being merged.
    • Delete feature branches once they are merged into main.

  2. Commit Messages:

    • Write clear and concise commit messages describing the changes made.
    • Use the imperative mood, e.g., "Add feature" not "Added feature" or "Adds feature".

  3. Conflict Resolution:

    • Resolve merge conflicts in your development branch before submitting a pull request.
    • Keep your branch updated with the latest changes from main to minimize conflicts.

  4. Code Review:

    • Actively participate in code review processes.
    • Reviewers should ensure code quality, functionality, and adherence to design principles.

Specs

  1. Status: MVP
  2. Architecture:
    1. Involved Platforms:
      1. GCP
        1. Stack:
          1. Cloud Run (europe-west3-docker.pkg.dev/...)
          2. BQ
  3. .env

ENV-Configs

NUM_STEPS=10 (todo :: UI :: DFS search of user input)

MODEL=gpt-3.5-turbo-16k-0613 (todo :: UI :: Dropdown of models)

Credentials

BQ_CLIENT_SECRETS={*****}

OPENAI_API_KEY=****

Endpoints

OPENAI_BASE_URL=https://api.openai.com/v1/chat/completions

Tables

ADJACENCY_MATRIX_DATASET_ID=graph_to_agent_adjacency_matrices MULTI_LAYERED_MATRIX_DATASET_ID=graph_to_agent_multi_layered_metrices ANSWER_CURATED_CHAT_COMPLETIONS=graph_to_agent_answer_curated_chat_completions CURATED_CHAT_COMPLETIONS=graph_to_agent_chat_completions RAW_CHAT_COMPLETIONS=graph_to_agent_raw_chat_completions GRAPH_DATASET_ID=graph_to_agent EDGES_TABLE=edges_table NODES_TABLE=nodes_table

Local Dirs

TEMP_RAW_CHAT_COMPLETIONS_DIR=temp_raw_chat_completions TEMP_MULTI_LAYERED_MATRIX_DIR=temp_multi_layered_matrix TEMP_CHECKPOINTS_GPT_CALLS=temp_checkpoints_gpt_calls LOG_DIR_LOCAL=./temp_log

Video-Agenda

  1. control + shift --> default :D
  2. control c + v
  3. explain multi select difference between custom function, agent and assistant
  4. modi msg passing
  5. how to read proposal
  6. explain roadmap
  7. explain interest in simulation
    1. law of physics connected with msg. passing weights/ the more msgs/ the more info one msg carries, the higher the gravity of a node
    2. conways game of life
    3. llama-index --> language corpus via edges
  8. explain different dimensions if somebody uses a pre-trained/ fine tuned model
  9. Explain the sheer infinite statistical possibilities, starting with the two layers
    1. explain the difference between the two layers
    2. distance matrix
    3. social interaction simulation by msg. protocols
    4. Agent behaviour with dynamic environment