datagemma-rag-27b-it-IMat-GGUF

Llama.cpp imatrix quantization of google/datagemma-rag-27b-it

Original Model: google/datagemma-rag-27b-it
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3750
IMatrix dataset: here


Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
datagemma-rag-27b-it.Q8_0.gguf Q8_0 28.94GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q6_K.gguf Q6_K 22.34GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q4_K.gguf Q4_K 16.65GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q3_K.gguf Q3_K 13.42GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q2_K.gguf Q2_K 10.45GB ✅ Available 🟢 IMatrix 📦 No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
datagemma-rag-27b-it.BF16/* BF16 54.46GB ✅ Available ⚪ Static ✂ Yes
datagemma-rag-27b-it.FP16/* F16 54.46GB ✅ Available ⚪ Static ✂ Yes
datagemma-rag-27b-it.Q8_0.gguf Q8_0 28.94GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q6_K.gguf Q6_K 22.34GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q5_K.gguf Q5_K 19.41GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q5_K_S.gguf Q5_K_S 18.88GB ✅ Available ⚪ Static 📦 No
datagemma-rag-27b-it.Q4_K.gguf Q4_K 16.65GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q4_K_S.gguf Q4_K_S 15.74GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ4_NL.gguf IQ4_NL 15.63GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ4_XS.gguf IQ4_XS 14.81GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q3_K.gguf Q3_K 13.42GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q3_K_L.gguf Q3_K_L 14.52GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q3_K_S.gguf Q3_K_S 12.17GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ3_M.gguf IQ3_M 12.45GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ3_S.gguf IQ3_S 12.17GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ3_XS.gguf IQ3_XS 11.55GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ3_XXS.gguf IQ3_XXS 10.75GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q2_K.gguf Q2_K 10.45GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.Q2_K_S.gguf Q2_K_S 9.72GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ2_M.gguf IQ2_M 9.40GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ2_S.gguf IQ2_S 8.65GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ2_XS.gguf IQ2_XS 8.40GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ2_XXS.gguf IQ2_XXS 7.63GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ1_M.gguf IQ1_M 6.69GB ✅ Available 🟢 IMatrix 📦 No
datagemma-rag-27b-it.IQ1_S.gguf IQ1_S 6.13GB ✅ Available 🟢 IMatrix 📦 No

Downloading using huggingface-cli

If you do not have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Download the specific file you want:

huggingface-cli download legraphista/datagemma-rag-27b-it-IMat-GGUF --include "datagemma-rag-27b-it.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/datagemma-rag-27b-it-IMat-GGUF --include "datagemma-rag-27b-it.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<bos><start_of_turn>user
{user_prompt}<end_of_turn>
<start_of_turn>model
{assistant_response}<end_of_turn>
<start_of_turn>user
{next_user_prompt}<end_of_turn>

Llama.cpp

llama.cpp/main -m datagemma-rag-27b-it.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"

FAQ

Why is the IMatrix not applied everywhere?

According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).

How do I merge a split GGUF?

  1. Make sure you have gguf-split available
  2. Locate your GGUF chunks folder (ex: datagemma-rag-27b-it.Q8_0)
  3. Run gguf-split --merge datagemma-rag-27b-it.Q8_0/datagemma-rag-27b-it.Q8_0-00001-of-XXXXX.gguf datagemma-rag-27b-it.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!

Downloads last month
158
GGUF
Model size
27.2B params
Architecture
gemma2

1-bit

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

16-bit

Inference Examples
Inference API (serverless) has been turned off for this model.

Model tree for legraphista/datagemma-rag-27b-it-IMat-GGUF

Base model

google/gemma-2-27b
Quantized
(14)
this model