Modelpublic

second-state/DeepSeek-V2-Lite-Chat-GGUF

sourceHugging Faceotherupdated 2y agoView on Hugging Face

6likes2.7kdownloads

Model Card

<div style="width: auto; margin-left: auto; margin-right: auto"> <img src="https://github.com/LlamaEdge/LlamaEdge/raw/dev/assets/logo.svg" style="width: 100%; min-width: 400px; display: block; margin: auto;"> </div> <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">

DeepSeek-V2-Lite-Chat-GGUF

Original Model

deepseek-ai/DeepSeek-V2-Lite-Chat

Run with LlamaEdge

—LlamaEdge version: v0.12.3 and above

—Prompt template

—Prompt type: deepseek-chat-2

—Prompt string

text

    <|begin_of_sentence|>{system_message}

    User: {user_message_1}

    Assistant: {assistant_message_1}<|end_of_sentence|>User: {user_message_2}

    Assistant:

—Context size: 128000

<!-- - Run as LlamaEdge service

bash

  wasmedge --dir .:. \
    --nn-preload default:GGML:AUTO:DeepSeek-V2-Lite-Chat-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template deepseek-chat-2 \
    --ctx-size 128000 \
    --model-name DeepSeek-V2-Lite-Chat

—Run as LlamaEdge command app

bash

  wasmedge --dir .:. \
    --nn-preload default:GGML:AUTO:DeepSeek-V2-Lite-Chat-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template deepseek-chat-2 \
    --ctx-size 128000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
DeepSeek-V2-Lite-Chat-Q2_K.gguf	Q2_K	2	6.43 GB	smallest, significant quality loss - not recommended for most purposes
DeepSeek-V2-Lite-Chat-Q3_K_L.gguf	Q3KL	3	8.46 GB	small, substantial quality loss
DeepSeek-V2-Lite-Chat-Q3_K_M.gguf	Q3KM	3	8.13 GB	very small, high quality loss
DeepSeek-V2-Lite-Chat-Q3_K_S.gguf	Q3KS	3	7.49 GB	very small, high quality loss
DeepSeek-V2-Lite-Chat-Q4_0.gguf	Q4_0	4	8.91 GB	legacy; small, very high quality loss - prefer using Q3KM
DeepSeek-V2-Lite-Chat-Q4_K_M.gguf	Q4KM	4	10.4 GB	medium, balanced quality - recommended
DeepSeek-V2-Lite-Chat-Q4_K_S.gguf	Q4KS	4	9.53 GB	small, greater quality loss
DeepSeek-V2-Lite-Chat-Q5_0.gguf	Q5_0	5	10.8 GB	legacy; medium, balanced quality - prefer using Q4KM
DeepSeek-V2-Lite-Chat-Q5_K_M.gguf	Q5KM	5	11.9 GB	large, very low quality loss - recommended
DeepSeek-V2-Lite-Chat-Q5_K_S.gguf	Q5KS	5	11.1 GB	large, low quality loss - recommended
DeepSeek-V2-Lite-Chat-Q6_K.gguf	Q6_K	6	14.1 GB	very large, extremely low quality loss
DeepSeek-V2-Lite-Chat-Q8_0.gguf	Q8_0	8	16.7 GB	very large, extremely low quality loss - not recommended
DeepSeek-V2-Lite-Chat-f16.gguf	f16	16	31.4 GB

Quatized with llama.cpp b3333