Edit model card

RoLlama2-7b-Instruct-IMat-GGUF

Llama.cpp imatrix quantization of RoLlama2-7b-Instruct-IMat-GGUF

Original Model: OpenLLM-Ro/RoLlama2-7b-Instruct
Original dtype: FP32 (float32)
Quantized by: llama.cpp b2998
IMatrix dataset: here

Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
RoLlama2-7b-Instruct.Q8_0.gguf Q8_0 7.16GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.Q6_K.gguf Q6_K 5.53GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.Q4_K.gguf Q4_K 4.08GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.Q3_K.gguf Q3_K 3.30GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.Q2_K.gguf Q2_K 2.53GB βœ… Available 🟒 Yes πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
RoLlama2-7b-Instruct.FP16.gguf F16 13.48GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.BF16.gguf BF16 13.48GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.Q5_K.gguf Q5_K 4.78GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.Q5_K_S.gguf Q5_K_S 4.65GB βœ… Available βšͺ No πŸ“¦ No
RoLlama2-7b-Instruct.Q4_K_S.gguf Q4_K_S 3.86GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.Q3_K_L.gguf Q3_K_L 3.60GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.Q3_K_S.gguf Q3_K_S 2.95GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.Q2_K_S.gguf Q2_K_S 2.32GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ4_NL.gguf IQ4_NL 3.83GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ4_XS.gguf IQ4_XS 3.62GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ3_M.gguf IQ3_M 3.11GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ3_S.gguf IQ3_S 2.95GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ3_XS.gguf IQ3_XS 2.80GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ3_XXS.gguf IQ3_XXS 2.59GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ2_M.gguf IQ2_M 2.36GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ2_S.gguf IQ2_S 2.20GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ2_XS.gguf IQ2_XS 2.03GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ2_XXS.gguf IQ2_XXS 1.85GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ1_M.gguf IQ1_M 1.65GB βœ… Available 🟒 Yes πŸ“¦ No
RoLlama2-7b-Instruct.IQ1_S.gguf IQ1_S 1.53GB βœ… Available 🟒 Yes πŸ“¦ No

Downloading using huggingface-cli

First, make sure you have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download legraphista/RoLlama2-7b-Instruct-IMat-GGUF --include "RoLlama2-7b-Instruct.Q8_0.gguf" --local-dir ./

If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download legraphista/RoLlama2-7b-Instruct-IMat-GGUF --include "RoLlama2-7b-Instruct.Q8_0/*" --local-dir RoLlama2-7b-Instruct.Q8_0
# see FAQ for merging GGUF's

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: RoLlama2-7b-Instruct.Q8_0)
  3. Run gguf-split --merge RoLlama2-7b-Instruct.Q8_0/RoLlama2-7b-Instruct.Q8_0-00001-of-XXXXX.gguf RoLlama2-7b-Instruct.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
2,201
GGUF
Model size
6.74B params
Architecture
llama

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/RoLlama2-7b-Instruct-IMat-GGUF

Quantized
(6)
this model

Collection including legraphista/RoLlama2-7b-Instruct-IMat-GGUF