Skip to content

Latest commit

 

History

History
54 lines (31 loc) · 1.77 KB

CHANGELOG.md

File metadata and controls

54 lines (31 loc) · 1.77 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

[2.0.1]

Changes

  • Structured LLM output with JSON Schema and GBNF Grammar Support for model type 'GGUF_CPU'

[2.0.0]

Changes

  • Introducing new model type: GGUF_CPU to support running gguf compiled models on CPU with LLM.js
  • Bumped up Llama.cpp worker to build commit dd047b4
  • Model Caching: Persisting model files in WebWorker's cache to avoid re-downloading model on every load. Thanks to PR from @felladrin.

Removes

  • Deprecating model types: LLAMA2, DOLLY_V2, GPT_2, GPT_J, GPT_NEO_X, MPT, REPLIT, STARCODER

[1.0.2]

Added

  • Latest GGUF format supported with latest updated llama2-cpp module.
  • Added context_size parameter for llama models.

Changed

  • Rebranded project to "LLM.JS".
  • Removed STACK_SIZE flag from build scripts.

[1.0.1]

Added

Changed

  • Added tokenizer_url parameter in GGML Wrapper for supporting LLaMa2.
  • Upgraded ggml version to 244776a

[1.0.0]

Added

  • ggml.js package
  • Added Model Support for Dollyv2, GPT2, GPT J, GPT Neo X, MPT, Replit, StarCoder
  • docsify documentation