Gemma 2
Collection
6 items
•
Updated
Llama.cpp imatrix quantization of google/shieldgemma-2b
Original Model: google/shieldgemma-2b
Original dtype: BF16
(bfloat16
)
Quantized by: llama.cpp b3525
IMatrix dataset: here
Status: ✅ Available
Link: here
Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
---|---|---|---|---|---|
shieldgemma-2b.Q8_0.gguf | Q8_0 | 2.78GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q6_K.gguf | Q6_K | 2.15GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q4_K.gguf | Q4_K | 1.71GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q3_K.gguf | Q3_K | 1.46GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q2_K.gguf | Q2_K | 1.23GB | ✅ Available | 🟢 IMatrix | 📦 No |
Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
---|---|---|---|---|---|
shieldgemma-2b.BF16.gguf | BF16 | 5.24GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.FP16.gguf | F16 | 5.24GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q8_0.gguf | Q8_0 | 2.78GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q6_K.gguf | Q6_K | 2.15GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q5_K.gguf | Q5_K | 1.92GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q5_K_S.gguf | Q5_K_S | 1.88GB | ✅ Available | ⚪ Static | 📦 No |
shieldgemma-2b.Q4_K.gguf | Q4_K | 1.71GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q4_K_S.gguf | Q4_K_S | 1.64GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ4_NL.gguf | IQ4_NL | 1.63GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ4_XS.gguf | IQ4_XS | 1.57GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q3_K.gguf | Q3_K | 1.46GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q3_K_L.gguf | Q3_K_L | 1.55GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q3_K_S.gguf | Q3_K_S | 1.36GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ3_M.gguf | IQ3_M | 1.39GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ3_S.gguf | IQ3_S | 1.36GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ3_XS.gguf | IQ3_XS | 1.31GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ3_XXS.gguf | IQ3_XXS | 1.18GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q2_K.gguf | Q2_K | 1.23GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.Q2_K_S.gguf | Q2_K_S | 1.17GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ2_M.gguf | IQ2_M | 1.09GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ2_S.gguf | IQ2_S | 1.03GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ2_XS.gguf | IQ2_XS | 1.00GB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ2_XXS.gguf | IQ2_XXS | 943.19MB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ1_M.gguf | IQ1_M | 873.80MB | ✅ Available | 🟢 IMatrix | 📦 No |
shieldgemma-2b.IQ1_S.gguf | IQ1_S | 832.16MB | ✅ Available | 🟢 IMatrix | 📦 No |
If you do not have hugginface-cli installed:
pip install -U "huggingface_hub[cli]"
Download the specific file you want:
huggingface-cli download legraphista/shieldgemma-2b-IMat-GGUF --include "shieldgemma-2b.Q8_0.gguf" --local-dir ./
If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:
huggingface-cli download legraphista/shieldgemma-2b-IMat-GGUF --include "shieldgemma-2b.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's
<bos>You are a policy expert trying to help determine whether a user prompt is in violation of the defined safety policies.
<start_of_turn>
Human Question: {next_user_prompt}
<end_of_turn>
Our safety principle is defined in the below:
* {guideline}
===
Does the human question violate the above principle? Your answer must start with 'Yes' or 'No'. And then walk through step by step to be sure we answer correctly.
llama.cpp/main -m shieldgemma-2b.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"
According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).
gguf-split
availablegguf-split
, navigate to https://github.com/ggerganov/llama.cpp/releasesgguf-split
shieldgemma-2b.Q8_0
)gguf-split --merge shieldgemma-2b.Q8_0/shieldgemma-2b.Q8_0-00001-of-XXXXX.gguf shieldgemma-2b.Q8_0.gguf
gguf-split
to the first chunk of the split.Got a suggestion? Ping me @legraphista!
Base model
google/shieldgemma-2b