Copyright (c) 2025 CommanderLake All rights reserved.

🦙 LM Stud – Local LLMs Minus the Lard

TL;DR
A WinForms chat client for llama.cpp, whisper.cpp and your questionable life choices.
Zero Electron. Zero telemetry. Zero regrets.

---

Features

☑️	What It Does
✅	Launches in milliseconds—your GPU blinks and it’s already chatting.
✅	Edit / regenerate / yeet messages.
✅	Shows the model’s “thinking” stream (fun for prompt nerds).
✅	Drag-drop files → instant code blocks.
✅	One-click Hugging Face search & download.
✅	Built-in Google Search + webpage fetch (super-visible setup below).
✅	Optional speech I/O with whisper.cpp—talk smack to your computer.
✅	Tiny memory footprint—smaller than RGB keyboard driver.
✅	Model API handler for remote endpoints.
✅	Dialectic mode with dual samplers for side-by-side debates.

---

Google Search – READ ME FIRST ⚠️

1)  Grab an API key
    https://console.cloud.google.com/apis/dashboard
    → new project → enable “Custom Search API” → copy the key.

2)  Create a Search Engine ID
    https://programmablesearchengine.google.com/controlpanel/overview
    → “Add” → “Search the entire web” → grab the cx ID.

3)  Paste both values in  Settings → Google Search Tool.
    Congrats—~100 free queries per day. Abuse responsibly.

---

Screenshots

Chat Tab	Settings Tab	Models Tab	Hugging Face Tab

---

Quick-Start Build

Build llama.cpp

1. Install Visual Studio with C# and C++ development tools
2. Clone https://github.com/ggml-org/llama.cpp or download the source code of the latest release
3. Open "x64 Native Tools Command Prompt for VS <version>", cd to the llama.cpp folder and run "mkdir build && cd build"
4. `cmake .. -DGGML_NATIVE=OFF -DGGML_BACKEND_DL=ON -DGGML_AVX2=ON -DGGML_BMI2=ON -DGGML_CUDA=ON -DGGML_CUDA_F16=ON -DLLAMA_CURL=OFF -DLLAMA_ALL_WARNINGS=OFF -DLLAMA_BUILD_TESTS=OFF -DLLAMA_BUILD_TOOLS=OFF -DLLAMA_BUILD_EXAMPLES=OFF -DLLAMA_BUILD_SERVER=OFF`
5. Open build\llama.cpp.sln with visual studio and build the Release version

Copy files

1. Clone https://github.com/CommanderLake/LMStud
2. Adjust "VC++ Directories" in the Stud project so the paths reflect where you cloned llama.cpp and whisper.cpp to
3. From build\bin\Release copy all .dll files to <LMStud solution folder>\LM Stud\bin\x64\Release
4. If you wish to use the Debug version follow the last 2 steps but replace "Release" with "Debug", i use Release for CUDA 10.2 and ReleaseCUDA12 for the CUDA 12.8 build
5. If you want whisper.cpp for voice input the build steps are similar to llama.cpp, clone https://github.com/ggml-org/whisper.cpp copy only whisper.dll to the "LM Stud\bin\x64\Release" folder or whatever your build configuration is

Set up curl for tools and model downloading

1. Set up vcpkg, step "1 - Set up vcpkg": https://learn.microsoft.com/en-us/vcpkg/get_started/get-started
2. From a new visual studio command prompt...
3. vcpkg install SDL2:x64-windows-static
4. vcpkg install curl[openssl]:x64-windows-static
5. Open LM Stud.sln in Visual Studio
6. Adjust "Include Directories" in "VC++ Directories" in the Stud project to your llama.cpp, whisper.cpp and vcpkg locations
7. Build LM Stud

---

Settings Cheat-Sheet

Section	Knobs & Dials
CPU Params / Batch	Generation threads, batch threads.
Common	Context size, GPU layers, temperature, tokens to generate.
Advanced	NUMA strategy, repeat penalty, top-k/p, batch size.
Voice	Model picker, wake word, VAD, frequency threshold, GPU toggle.
Tools	Enable Google Search (API key + cx), enable webpage fetch.

---

Recent Releases

• R37 – Added dual samplers for dialectic mode and switched the UI font to Segoe UI Symbol.
• R36 – Added tooltip for the generation delay setting.
• R35 – Improved generation lock so Generate no longer waits.
• R34 – Introduced a model API handler.

---

License

No license granted. All rights reserved.
Viewing and forking on GitHub are permitted under GitHub’s Terms of Service.
No other rights are granted; any use, redistribution, or modification outside GitHub requires prior written permission.

Disclaimer

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE and NON-INFRINGEMENT, TO THE MAXIMUM EXTENT PERMITTED BY LAW.