CoolFace
Modelpublic

second-state/Octopus-v2-GGUF

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
6likes682downloads
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 -->

Octopus-v2-2B-GGUF

Original Model

NexaAIDev/Octopus-v2

Run with LlamaEdge

  • LlamaEdge version: v0.8.1 and above
  • Prompt template
  • Prompt type: octopus
  • Prompt string
console
      {system_prompt}\n\nQuery: {input_text} \n\nResponse:
  • Context size: 2048
  • Run as LlamaEdge service
bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Octopus-v2-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template octopus \
    --ctx-size 2048 \
    --model-name octopus-v2

Example of a user request in json format:

json
  {
      "messages": [
          {
              "role": "system",
              "content": "Below is the query from the users, please call the correct function and generate the parameters to call the function."
          },
          {
              "role": "user",
              "content": "Take a selfie for me with front camera"
          }
      ],
      "model": "octopus-v2",
      "stream": false
  }

Quantized GGUF Models

NameQuant methodBitsSizeUse case
Octopus-v2-Q2_K.ggufQ2_K21.16 GBsmallest, significant quality loss - not recommended for most purposes
Octopus-v2-Q3_K_L.ggufQ3KL31.47 GBsmall, substantial quality loss
Octopus-v2-Q3_K_M.ggufQ3KM31.38 GBvery small, high quality loss
Octopus-v2-Q3_K_S.ggufQ3KS31.29 GBvery small, high quality loss
Octopus-v2-Q4_0.ggufQ4_041.55 GBlegacy; small, very high quality loss - prefer using Q3KM
Octopus-v2-Q4_K_M.ggufQ4KM41.63 GBmedium, balanced quality - recommended
Octopus-v2-Q4_K_S.ggufQ4KS41.56 GBsmall, greater quality loss
Octopus-v2-Q5_0.ggufQ5_051.8 GBlegacy; medium, balanced quality - prefer using Q4KM
Octopus-v2-Q5_K_M.ggufQ5KM51.84 GBlarge, very low quality loss - recommended
Octopus-v2-Q5_K_S.ggufQ5KS51.8 GBlarge, low quality loss - recommended
Octopus-v2-Q6_K.ggufQ6_K62.06 GBvery large, extremely low quality loss
Octopus-v2-Q8_0.ggufQ8_082.67 GBvery large, extremely low quality loss - not recommended
Octopus-v2-f16.gguff161610 GB

Quantized with llama.cpp b2589