---

Breeze-7B-Instruct-v1_0-GGUF

Original Model

MediaTek-Research/Breeze-7B-Instruct-v1_0

Run with LlamaEdge

• LlamaEdge version: v0.14.2 and above

• Prompt template

  • Prompt type: mediatek-breeze

  • Prompt string

    <s>{system_message}  [INST] {user_message_1} [/INST] {assistant_message_1} [INST] {user_message_2} [/INST]
    

• Context size: 8000

• Run as LlamaEdge service

  • Chat

    wasmedge --dir .:. --nn-preload default:GGML:AUTO:Breeze-7B-Instruct-v1_0-Q5_K_M.gguf \
      llama-api-server.wasm \
      --prompt-template mediatek-breeze \
      --ctx-size 8000 \
      --model-name breeze-7b
    

• Run as LlamaEdge command app

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Breeze-7B-Instruct-v1_0-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template mediatek-breeze \
    --ctx-size 8000
  

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Breeze-7B-Instruct-v1_0-Q2_K.gguf	Q2_K	2	2.86 GB	smallest, significant quality loss - not recommended for most purposes
Breeze-7B-Instruct-v1_0-Q3_K_L.gguf	Q3_K_L	3	3.98 GB	small, substantial quality loss
Breeze-7B-Instruct-v1_0-Q3_K_M.gguf	Q3_K_M	3	3.67 GB	very small, high quality loss
Breeze-7B-Instruct-v1_0-Q3_K_S.gguf	Q3_K_S	3	3.32 GB	very small, high quality loss
Breeze-7B-Instruct-v1_0-Q4_0.gguf	Q4_0	4	4.28 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Breeze-7B-Instruct-v1_0-Q4_K_M.gguf	Q4_K_M	4	4.54 GB	medium, balanced quality - recommended
Breeze-7B-Instruct-v1_0-Q4_K_S.gguf	Q4_K_S	4	4.31 GB	small, greater quality loss
Breeze-7B-Instruct-v1_0-Q5_0.gguf	Q5_0	5	5.18 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Breeze-7B-Instruct-v1_0-Q5_K_M.gguf	Q5_K_M	5	5.32 GB	large, very low quality loss - recommended
Breeze-7B-Instruct-v1_0-Q5_K_S.gguf	Q5_K_S	5	5.18 GB	large, low quality loss - recommended
Breeze-7B-Instruct-v1_0-Q6_K.gguf	Q6_K	6	6.14 GB	very large, extremely low quality loss
Breeze-7B-Instruct-v1_0-Q8_0.gguf	Q8_0	8	7.96 GB	very large, extremely low quality loss - not recommended
Breeze-7B-Instruct-v1_0-f16.gguf	f16	16	15.0 GB

Quantized with llama.cpp b3613