Narok
Full Changelog: v0.2.9...v0.2.9
Here is a release note template tailored for GitHub for Malama v0.2.9:
🚀 Malama v0.2.9
Malama v0.2.9 introduces per-session parameter tuning, pre-flight context budgeting, conversation export capabilities, and test suite expansions to elevate your native local LLM chat experience on Linux!
🌟 What's New in v0.2.9
🎛️ Per-Session Parameter Tuning & System Prompts
- Dynamic Inference Parameters: Customize
temperature,top_p,top_k,repeat_penalty, andnum_ctxindependently for each conversation thread. - System Prompt Overrides: Integrated modal panel to inject custom system instructions per session before dispatching prompts to Ollama.
📊 Pre-Flight Token Budget & Attachment Estimator
- Real-time Budgeting: Instant calculation of prompt + attachment context size before submitting a request.
- Context Overflow Warnings: Visual warning indicator alerts you when your total context payload approaches or exceeds the configured
num_ctxboundary.
💾 Native Conversation Export Subsystem (ExportEngine)
- Export active chat sessions with formatting preserved into:
- Markdown (
.md) - JSON (
.json) - Plain Text (
.txt)
⚙️ Engine & Build Enhancements
- Bumped CMake project version to
0.2.9. - Expanded Catch2 v3 automated test coverage for parameter serialization and export formatting.
- Containerized Arch Linux CI/CD release pipeline for standardized AppImage assembly.
📦 Installation & Usage
AppImage (Recommended)
Download Malama-v0.2.9-x86_64.AppImage from the assets below and make it executable:
chmod +x Malama-v0.2.9-x86_64.AppImage
./Malama-v0.2.9-x86_64.AppImage
Build from Source
git clone https://github.com/Magpiny/malama.git
cd malama
cmake -B build -G Ninja -DCMAKE_BUILD_TYPE=Release
cmake --build build --parallel $(nproc)
./build/malama
Note: Ensure your local [Ollama](https://ollama.com/) service is active (
ollama serve).