CoolFace
Modelpublic

second-state/CodeGemma-7b-it-GGUF

sourceHugging Facegemmaupdated 2y agoView on Hugging Face
2likes380downloads
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 -->

CodeGemma-7b-it-GGUF

Original Model

google/codegemma-7b-it

Run with LlamaEdge

  • —LlamaEdge version: v0.8.1 and above
  • —Prompt template
  • —Prompt type: gemma-instruct
  • —Prompt string
text
    <bos><start_of_turn>user
    {user_message}<end_of_turn>
    <start_of_turn>model
    {model_message}<end_of_turn>model
  • —Context size: 3072
  • —Run as LlamaEdge service
bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:codegemma-7b-it-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template gemma-instruct \
    --ctx-size 3072 \
    --model-name codegemma-7b

<!-- - Run as LlamaEdge command app

bash
  wasmedge --dir .:. --nn-preload default:GGML:AUTO:gemma-7b-it-Q5_K_M.gguf llama-chat.wasm -p gemma-instruct -c 4096

Quantized GGUF Models

NameQuant methodBitsSizeUse case
codegemma-7b-it-Q2_K.ggufQ2_K23.48 GBsmallest, significant quality loss - not recommended for most purposes
codegemma-7b-it-Q3_K_L.ggufQ3KL34.71 GBsmall, substantial quality loss
codegemma-7b-it-Q3_K_M.ggufQ3KM34.37 GBvery small, high quality loss
codegemma-7b-it-Q3_K_S.ggufQ3KS33.98 GBvery small, high quality loss
codegemma-7b-it-Q4_0.ggufQ4_045.01 GBlegacy; small, very high quality loss - prefer using Q3KM
codegemma-7b-it-Q4_K_M.ggufQ4KM45.33 GBmedium, balanced quality - recommended
codegemma-7b-it-Q4_K_S.ggufQ4KS45.05 GBsmall, greater quality loss
codegemma-7b-it-Q5_0.ggufQ5_055.98 GBlegacy; medium, balanced quality - prefer using Q4KM
codegemma-7b-it-Q5_K_M.ggufQ5KM56.14 GBlarge, very low quality loss - recommended
codegemma-7b-it-Q5_K_S.ggufQ5KS55.98 GBlarge, low quality loss - recommended
codegemma-7b-it-Q6_K.ggufQ6_K67.01 GBvery large, extremely low quality loss
codegemma-7b-it-Q8_0.ggufQ8_089.08 GBvery large, extremely low quality loss - not recommended
codegemma-7b-it-f16.gguff161617.1 GB

Quantized with llama.cpp b2589