You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
feat: update llama.cpp dependency and improve model loading
- Remove direct llama.cpp Swift package dependency in favor of
pre-built xcframework
- Update LlamaFramework binary to latest version b7681
- Replace llama_load_model_from_file with llama_model_load_from_file
- Replace llama_new_context_with_model with llama_init_from_model
- Add vocab property to LlamaModel class and use llama_model_get_vocab
- Update token handling to use vocab instead of model for token operations
- Replace llama_n_ctx_train with llama_model_n_ctx_train
- Replace llama_token_is_eog with llama_vocab_is_eog
- Update llama_token_to_piece calls to use vocab parameter
- Replace llama_free_model with llama_model_free
- Add proper memory management with llama_get_memory and llama_memory_clear
- Update README.md package version from 0.4.0 to 0.4.2
- Add Vocab type alias in TypeAlias.swift
- Remove Package.resolved files
- Update project build configurations and package references
BREAKING CHANGE: The llama.cpp dependency is now provided as a
pre-built xcframework instead of source dependency.