cactus

What's Changed

• Kernel optimisations by @HenryNdubuaku in https://github.com/cactus-compute/cactus/pull/397
• Improve INT4 by @ncylich and @jrajala6 in https://github.com/cactus-compute/cactus/pull/343
• add einops dependency to requirements by @jakmro in https://github.com/cactus-compute/cactus/pull/371
• Add language parameter support for Whisper transcription by @rshemet in https://gith...

What's Changed

• Brew setup @HenryNdubuaku
• Cactus auth @HenryNdubuaku
• Hybrid inference by the cactus team
• Karen/vlm fix by @kar-m in https://github.com/cactus-compute/cactus/pull/311
• fixed moonshine state resetting and gemma3 4b layernorm loading by @kar-m in https://github.com/cactus-compute/cactus/pull/317
• fix: LFM2 multiple tool calls by @mhayes853 in https://github.com/ca...

What's Changed

• Kernel Optimisations & advanced quantisation by @HenryNdubuaku
• Moonshine by @kar-m
• HF publish by @jakmro
• Streaming API by @jakmro
• Linux ARM support by @ncylich
• Stop generation on model end token by @Ayan9074
• i8MM runtime detection @mhayes853

FFI Note: This break API

What's Changed

• Groupwise quantisation by @HenryNdubuaku
• Speech-To-Text streaming by @jakmro
• KV Quntisation by @HenryNdubuaku
• Evals by @justinl66 @ParkiratS
• INT4 support by @HenryNdubuaku
• Rust bindings by @mrsarac

Bindings: Please check Cactus FFIs again @jakmro @mrsarac @mhayes853

What's Changed

• Cactus index by @jakmro
• Function Gemma by @HenryNdubuaku
• Perplexity eval to model by @ParkiratS
• Tool refactor by @rshemet
• F16 kernel updates by @KayaanT
• Multi-turn VLM conversation continuity by @HenryNdubuaku
• Bugfixes by @ammesatyajit

Instruction for bindings @mhayes853

• Should easily replace v1.3 without headaches