llama_cpp_dart 0.2.1 copy "llama_cpp_dart: ^0.2.1" to clipboard
llama_cpp_dart: ^0.2.1 copied to clipboard

Published 2 days ago
SDKFlutter
PlatformAndroidiOSLinuxmacOSWindows
→

Metadata

Dart binding for llama.cpp --- high level wrappers for both Dart and Flutter

More...

0.2.1 #

  • Android: Added OpenCL support for GPU acceleration (#91).
  • Vision:
    • Fixed crash in mtmd context disposal.
    • Stable Qwen3-VL support.
  • Chat: Added experimental support for Qwen3-VL chat format (_exportQwen3Jinja).
  • Fixes:
    • Improved logging initialization (#88).
    • Fixed stream processing crash in chat.
  • Core: Updated llama.cpp submodule.

0.2.0 #

  • llama.cpp 4ffc47cb2001e7d523f9ff525335bbe34b1a2858
  • Memory Safety: No more pointer being freed was not allocated crashes.
  • UTF-8 Safety: Emojis and foreign languages won't break generation.
  • Context Management: You can Append, Clear, or Save the brain.
  • Multi-Tenancy: You can handle multiple users (Slots) if you need to.
  • breakign changes: Old: Llama(path, modelParams, contextParams, samplerParams, verbose) New: Llama(path, modelParams: ..., contextParams: ..., samplerParams: ..., verbose: ...)
  • Parameter Serialization: ContextParams and ModelParams JSON serialization has changed. Enums (like LlamaRopeScalingType) now store their specific C-API integer values instead of Dart list indices. Old JSON configs may need migration.
  • Sampler Standardization: SamplerParams has been refactored to strictly match llama.cpp. Non-standard fields (e.g., xtcTemperature, topPKeep) have been renamed or removed.
  • RPC Removed: Removed rpcServers from ModelParams as it is no longer supported in the core struct.

0.1.2+1 #

  • forgot to update version

0.1.2 #

  • removed flash_attn llama_context_default_params
  • removed softmax
  • updated llama.cpp to b8595b16e

0.1.1 #

  • State load / save
  • llama.cpp 25ff6f7659f6a5c47d6a73eada5813f0495331f0
  • harmony prompting syntax
  • isolate has vision and verbose support
  • mcp server / agent
  • scope generation stopping

0.1.0 #

  • Multimodal support - vision

0.0.9 #

  • Major internal refactoring to improve code organization and maintainability
  • Fixed critical bug where subsequent prompts would fail due to batch seq_id memory management
  • Improved position tracking for continuous conversation support
  • Enhanced error handling and debugging capabilities
  • Added foundation for future chat optimization features
  • Breaking change: Internal API restructuring (public API remains stable)

0.0.8 #

  • disabled llava
  • compatible with llama.cpp 42ae10bb
  • add typed_isolate
  • removed llama processor

0.0.7 #

  • updated binding
  • performance imporvement and bugs fix

0.0.6 #

  • added initial support to load lora
  • dart cli example
  • fixed #3 by @danemadsen

0.0.5 #

  • removed assets defination
  • added static property Llama.libraryPath to set library path, in order to support linux and other platforms

0.0.4 #

  • ModelParams disabled options splitsMode, tensorSplit and metadataOverride

0.0.3 #

  • LlamaProcessor now take context and model parameters

0.0.2 #

  • refactored code to follow dart package structure

0.0.1 #

  • TODO: Describe initial release.

← Metadata

68
likes
130
points
224
downloads

Publisher

unverified uploader

Weekly Downloads

Metadata

Dart binding for llama.cpp --- high level wrappers for both Dart and Flutter

Repository (GitHub)
View/report issues

Documentation

API reference

License

MIT (license)

Dependencies

ffi, flutter, image, typed_isolate, uuid

More

Packages that depend on llama_cpp_dart

Back