💫 Community Model DeepSeek Coder V2 Lite Instruct by DeepSeek 👾 LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord . Model creator: DeepSeek Original model : DeepSeek Coder V2 Lite Instruct GGUF quantization: provided by bartowski based on llama.cpp release b3166 Model Settings: Requires LM Studio 0.2.25, update can be downloaded from here: https://lmstudio.ai Flash attention MUST be disabled for this model to work. Model Summary: This is a brand new Mixture of Export (MoE) model from DeepSeek, specializing in coding instructions. This model performs well across a series of coding benchmarks and should be used for both instruction following and code completion. Prompt template: The best performing template is Deepseek Coder preset in your LM Studio. This will format the prompt as follows: The "official" template seems to tend towards generating Chinese, however if you'd like to use it you can set it up by choosing the LM Studio Blank Preset preset in your LM Studio and then: Set your User Message Prefix to User: Set your User Message Suffix to \n\nAssistant: This will format the prompt as follows:…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy