CoolFace
Modelpublic

second-state/Mistral-7B-Instruct-v0.2-GGUF

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
1likes2.1kdownloads
Model Card

<!-- header start --> <!-- 200823 --> <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;"> <!-- header end -->

Mistral-7B-Instruct-v0.2-GGUF

Original Model

mistralai/Mistral-7B-Instruct-v0.2

Run with LlamaEdge

  • —LlamaEdge version: v0.2.8 and above
  • —Prompt template
  • —Prompt type: mistral-instruct
  • —Prompt string
text
    <s>[INST] {prompt} [/INST]
  • —Context size: 4096
  • —Run as LlamaEdge service
bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Mistral-7B-Instruct-v0.2-Q5_K_M.gguf llama-api-server.wasm -p mistral-instruct
  • —Run as LlamaEdge command app
bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Mistral-7B-Instruct-v0.2-Q5_K_M.gguf llama-chat.wasm -p mistral-instruct

Quantized GGUF Models

NameQuant methodBitsSizeUse case
Mistral-7B-Instruct-v0.2-Q2_K.ggufQ2_K23.08 GBsmallest, significant quality loss - not recommended for most purposes
Mistral-7B-Instruct-v0.2-Q3_K_L.ggufQ3KL33.82 GBsmall, substantial quality loss
Mistral-7B-Instruct-v0.2-Q3_K_M.ggufQ3KM33.52 GBvery small, high quality loss
Mistral-7B-Instruct-v0.2-Q3_K_S.ggufQ3KS33.16 GBvery small, high quality loss
Mistral-7B-Instruct-v0.2-Q4_0.ggufQ4_044.11 GBlegacy; small, very high quality loss - prefer using Q3KM
Mistral-7B-Instruct-v0.2-Q4_K_M.ggufQ4KM44.37 GBmedium, balanced quality - recommended
Mistral-7B-Instruct-v0.2-Q4_K_S.ggufQ4KS44.14 GBsmall, greater quality loss
Mistral-7B-Instruct-v0.2-Q5_0.ggufQ5_055 GBlegacy; medium, balanced quality - prefer using Q4KM
Mistral-7B-Instruct-v0.2-Q5_K_M.ggufQ5KM55.13 GBlarge, very low quality loss - recommended
Mistral-7B-Instruct-v0.2-Q5_K_S.ggufQ5KS55 GBlarge, low quality loss - recommended
Mistral-7B-Instruct-v0.2-Q6_K.ggufQ6_K65.94 GBvery large, extremely low quality loss
Mistral-7B-Instruct-v0.2-Q8_0.ggufQ8_087.7 GBvery large, extremely low quality loss - not recommended