Skip to content
View seonglae's full-sized avatar
:electron:
Knowledge absorbing
:electron:
Knowledge absorbing

Highlights

  • Pro

Organizations

@seongland @texonom

Block or report seonglae

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse

Pinned Loading

  1. CorrSteer CorrSteer Public

    CorrSteer: Generation-Time LLM Steering via Correlated Sparse Autoencoder Features

    Python 4 2

  2. intuiter intuiter Public

    Globally available Vim like productivity application with a fancy mouse control. Home row keyboard mapping makes typing innovatively fast in everywhere

    AutoHotkey 44 3

  3. FaithfulSAE FaithfulSAE Public

    Unlock the internal features of LLMs: FaithfulSAE trains sparse autoencoders only on synthetic data from the model itself, eliminating fake features and external data dependence for reproducible, r…

    Python 5

  4. llama2gptq llama2gptq Public

    Chat to LLaMa 2 that also provides responses with reference documents over vector database. Locally available model using GPTQ 4bit quantization.

    Python 31

  5. to-smooth to-smooth Public

    Chaikin's smoothing algorithm extended to a multidimensional library

    TypeScript 14 1

  6. texonom/notion-node texonom/notion-node Public

    Texonom's packages monorepo for Notion API & Utils for custom web page and data exporting

    TypeScript 7