Release v2.5.0

🚀 Features

• VisionOS support (PR: #299)
• Add support for Gemma 3 and Phi 4 models (PR: #327)
• Fix Android support for older devices (use ARMv8-A instead of ARMv8.4-A) (PR: #325)
• Upgrade LlamaLib to v1.2.4 (llama.cpp b4969) (PR: #325)
• Default number of predicted tokens (num_predict) to infinity (-1) (PR: #328)

Release v2.4.2

🚀 Features

• Integrate DeepSeek models (PR: #312)
• Update LlamaLib to v1.2.3 (llama.cpp b4688) (PR: #312)
• Drop CUDA 11.7.1 support (PR: #312)
• Add warm-up function for provided prompt (PR: #301)
• Add documentation in Unity tooltips (PR: #302)

🐛 Fixes

• Fix code signing on iOS (PR: #298)
• Persist debug mode and use of extras to the build (PR: #304)
• Fix dependency resolution for full CUDA and vulkan architectures (PR: #313)

Release v2.4.1

🚀 Features

• Static library linking on mobile (fixes iOS signing) (PR: #289)

🐛 Fixes

• Fix support for extras (flash attention, iQ quants) (PR: #292)

Release v2.4.0

🚀 Features

• iOS deployment (PR: #267)
• Improve building process (PR: #282)
• Add structured output / function calling sample (PR: #281)
• Update LlamaLib to v1.2.0 (llama.cpp b4218) (PR: #283)

🐛 Fixes

• Clear temp build directory before building (PR: #278)

📦 General

• Remove support for extras (flash attention, iQ quants) (PR: #284)
• remove support for LLM base prompt (PR: #285)

Release v2.3.0

🚀 Features

• Implement Retrieval Augmented Generation (RAG) in LLMUnity (PR: #246)

🐛 Fixes

• Fixed build conflict, endless import of resources. (PR: #266)

Release v2.2.4

🚀 Features

• Add Phi-3.5 and Llama 3.2 models (PR: #255)
• Speedup LLMCharacter warmup (PR: #257)

🐛 Fixes

• Fix handling of incomplete requests (PR: #251)
• Fix Unity locking of DLLs during cross-platform build (PR: #252)
• Allow spaces in lora paths (PR: #254)

📦 General

• Set default context size to 8192 and allow to adjust with a UI slider (PR: #258)

Release v2.2.3

🚀 Features

• LlamaLib v1.1.12: SSL certificate & API key for server, Support more AMD GPUs (PR: #241)
• Server security with API key and SSL (PR: #238)
• Show server command for easier deployment (PR #239)

🐛 Fixes

• Fix multiple LLM crash on Windows (PR: #242)
• Exclude system prompt from saving of chat history (PR: #240)

Release v2.2.2

🚀 Features

• Allow to set the LLMCharacter slot (PR: #231)

🐛 Fixes

• fix adding grammar from StreamingAssets (PR: #229)
• fix library setup restart when interrupted (PR: #232)
• Remove unnecessary Android linking in IL2CPP builds (PR: #233)

Release v2.2.1

🐛 Fixes

• Fix naming showing full path when loading model (PR: #224)
• Fix parallel prompts (PR: #226)

Release v2.2.0

🚀 Features

• Update to latest llama.cpp (b3617) (PR: #210)
• Integrate Llama 3.1 and Gemma2 models in model dropdown
• Implement embedding and lora adapter functionality (PR: #210)
• Read context length and warn if it is very large (PR: #211)
• Setup allowing to use extra features: flash attention and IQ quants (PR: #216)
• Allow HTTP request retries for remote server (PR: #217)
• Allow to set lora weights at startup, add unit test (PR: #219)
• allow relative StreamingAssets paths for models (PR: #221)

🐛 Fixes

• Fix set template for remote setup (PR: #208)
• fix crash when stopping scene before LLM creation (PR: #214)

📦 General

• Documentation/point to gguf format for lora (PR: #215)