CoolFace
Modelpublic

second-state/StarCoder2-7B-GGUF

sourceHugging Facebigcode-openrail-mupdated 3y agoView on Hugging Face
25likes10kdownloads
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 -->

StarCoder2-7B-GGUF

Original Model

bigcode/starcoder2-7b

Run with LlamaEdge

  • —LlamaEdge version: coming soon
  • —Context size: 4608

Quantized GGUF Models

NameQuant methodBitsSizeUse case
starcoder2-7b-Q2_K.ggufQ2_K22.72 GBsmallest, significant quality loss - not recommended for most purposes
starcoder2-7b-Q3_K_L.ggufQ3KL33.99 GBsmall, substantial quality loss
starcoder2-7b-Q3_K_M.ggufQ3KM33.59 GBvery small, high quality loss
starcoder2-7b-Q3_K_S.ggufQ3KS33.09 GBvery small, high quality loss
starcoder2-7b-Q4_0.ggufQ4_044.04 GBlegacy; small, very high quality loss - prefer using Q3KM
starcoder2-7b-Q4_K_M.ggufQ4KM44.4 GBmedium, balanced quality - recommended
starcoder2-7b-Q4_K_S.ggufQ4KS44.13 GBsmall, greater quality loss
starcoder2-7b-Q5_0.ggufQ5_054.94 GBlegacy; medium, balanced quality - prefer using Q4KM
starcoder2-7b-Q5_K_M.ggufQ5KM55.12 GBlarge, very low quality loss - recommended
starcoder2-7b-Q5_K_S.ggufQ5KS54.94 GBlarge, low quality loss - recommended
starcoder2-7b-Q6_K.ggufQ6_K65.89 GBvery large, extremely low quality loss
starcoder2-7b-Q8_0.ggufQ8_087.63 GBvery large, extremely low quality loss - not recommended

Quantized with llama.cpp b2308