v1.7.2

Highlights

• models used in README.md and tests are updated, as the models in the examples were too outdated
• updated xcframework

Full Changelog: v1.7.1...v1.7.2

v1.7.1

Highlight

• update xcframework

Full Changelog: v1.7.0...v1.7.1

v1.7.0

What's Changed

• improve performance, use actor by @eastriverlee in #45

Full Changelog: v1.6.2...v1.7.0

v1.6.2

Highlight

• add gemma template for google gemma models
• fix special token parsing

Full Changelog: v1.6.1...v1.6.2

v1.6.1

Highlight

• fix and simplify decode method #39
• remove redundant decode and decodeOnly methods

Full Changelog: v1.6.0...v1.6.1

v1.6.0

Highlight

• replaced deprecated llama.cpp APIs
• since llama.cpp doesn't have Package.swift anymore and build llama.xcframework on the fly(related llama.cpp commit), I made update.sh that updates llama.xcframework to the latest version, and replaced LLM.swift's dependency. Thus, pinned branch will no longer be updated
• fixed memory leak by calling llama_batch_free #43
• use Swift Test instead of XCTest, removing power assert dependency
• init is now optional, allowing users to check if initialized correctly instead of crashing instantly

Full Changelog: v1.5.5...v1.6.0

v1.5.5

What's Changed

• Update Package.swift by @0xjorgev in #38

New Contributors

• @0xjorgev made their first contribution in #38

Full Changelog: v1.5.4...v1.5.5

v1.5.4

What's Changed

• Support Xcode 16 beta 3 by @garfbargle in #29
• migrated to latest version of llama.cpp (2fb9267) by @garfbargle in #31
• sampling v2 update by @eastriverlee in #37

New Contributors

• @garfbargle made their first contribution in #29

Full Changelog: v1.5.3...v1.5.4

v1.5.3

What's Changed

• deal with API change by @eastriverlee in #27

Full Changelog: v1.5.2...v1.5.3

v1.5.2

Highlight

fixed failing tests #23

Full Changelog: v1.5.1...v1.5.2