-
Notifications
You must be signed in to change notification settings - Fork 469
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
rocm 7.14 introduced gfx1103 support but bundled llama.cpp gpu has no rocm backend
bugSomething isn't workingSomething isn't workingengine::llamacppllama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)llama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)runtime::rocmAMD ROCm runtimeAMD ROCm runtimeStatus: Open.#3230 In lemonade-sdk/lemonade;Lemonade Gemma4 leaking system prompt between different consumers
bugSomething isn't workingSomething isn't workingengine::llamacppllama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)llama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)Status: Open.#3228 In lemonade-sdk/lemonade;GUI3: provide an option for pure CLI args input
enhancementNew feature or requestNew feature or requestStatus: Open.#3227 In lemonade-sdk/lemonade;Problematic load/save semantics for args in the server
area::apiHTTP REST API surface and route handlersHTTP REST API surface and route handlersbugSomething isn't workingSomething isn't workingStatus: Open.#3226 In lemonade-sdk/lemonade;ctx-size auto-tune miscalculates VRAM headroom with multiple models on separate GPUs, silently falls back to 4096
bugSomething isn't workingSomething isn't workingengine::llamacppllama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)llama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)runtime::rocmAMD ROCm runtimeAMD ROCm runtimeStatus: Open.#3224 In lemonade-sdk/lemonade;Relay cloud providers that speak the Anthropic Messages wire format through /v1/messages
area::apiHTTP REST API surface and route handlersHTTP REST API surface and route handlersenhancementNew feature or requestNew feature or requestStatus: Open.#3221 In lemonade-sdk/lemonade;Support cloud providers with a non-standard auth header name/prefix
area::apiHTTP REST API surface and route handlersHTTP REST API surface and route handlersenhancementNew feature or requestNew feature or requestStatus: Open.#3220 In lemonade-sdk/lemonade;Web UI: API key entered in Settings has no effect until a manual page reload — looks like a broken Save button
bugSomething isn't workingSomething isn't workingStatus: Open.#3216 In lemonade-sdk/lemonade;Validating large model hashes stalls other operations
bugSomething isn't workingSomething isn't workingStatus: Open.#3214 In lemonade-sdk/lemonade;ROCm backend does not detect GPU on Windows when AMD driver ships a newer HIP runtime than TheRock
bugSomething isn't workingSomething isn't workingengine::llamacppllama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)llama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)runtime::rocmAMD ROCm runtimeAMD ROCm runtimeStatus: Open.#3213 In lemonade-sdk/lemonade;llamacppbackend install fails with "Failed to extract archive" for all backends (cpu/rocm/vulkan) on WindowsbugSomething isn't workingSomething isn't workingengine::llamacppllama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)llama.cpp backend (LlamaCppServer); GPU/CPU LLM inference (Vulkan, ROCm, Metal)Status: Open.#3207 In lemonade-sdk/lemonade;developer-setup link gives 404
documentationImprovements or additions to documentationImprovements or additions to documentationStatus: Open.#3206 In lemonade-sdk/lemonade;