README.md

metadata

base_model: OpenLLM-Ro/RoLlama2-7b-Chat
inference: false
language:
  - ro
library_name: gguf
license: cc-by-nc-4.0
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - quantized
  - GGUF
  - imatrix
  - quantization

RoLlama2-7b-Chat-IMat-GGUF

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

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

Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
RoLlama2-7b-Chat.Q8_0.gguf	Q8_0	7.16GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.Q6_K.gguf	Q6_K	5.53GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.Q4_K.gguf	Q4_K	4.08GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.Q3_K.gguf	Q3_K	3.30GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.Q2_K.gguf	Q2_K	2.53GB	✅ Available	Yes	📦 No

All Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
RoLlama2-7b-Chat.FP16.gguf	F16	13.48GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.BF16.gguf	BF16	13.48GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.Q5_K.gguf	Q5_K	4.78GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.Q5_K_S.gguf	Q5_K_S	4.65GB	✅ Available	No	📦 No
RoLlama2-7b-Chat.Q4_K_S.gguf	Q4_K_S	3.86GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.Q3_K_L.gguf	Q3_K_L	3.60GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.Q3_K_S.gguf	Q3_K_S	2.95GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.Q2_K_S.gguf	Q2_K_S	2.32GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.IQ4_NL.gguf	IQ4_NL	3.83GB	✅ Available	Yes	📦 No
RoLlama2-7b-Chat.IQ4_XS	IQ4_XS	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ3_M	IQ3_M	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ3_S	IQ3_S	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ3_XS	IQ3_XS	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ3_XXS	IQ3_XXS	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ2_M	IQ2_M	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ2_S	IQ2_S	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ2_XS	IQ2_XS	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ2_XXS	IQ2_XXS	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ1_M	IQ1_M	-	⏳ Processing	Yes	-
RoLlama2-7b-Chat.IQ1_S	IQ1_S	-	⏳ Processing	Yes	-

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-Chat-IMat-GGUF --include "RoLlama2-7b-Chat.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-Chat-IMat-GGUF --include "RoLlama2-7b-Chat.Q8_0/*" --local-dir RoLlama2-7b-Chat.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
   • To get hold of gguf-split, navigate to https://github.com/ggerganov/llama.cpp/releases
   • Download the appropriate zip for your system from the latest release
   • Unzip the archive and you should be able to find gguf-split
2. Locate your GGUF chunks folder (ex: RoLlama2-7b-Chat.Q8_0)
3. Run gguf-split --merge RoLlama2-7b-Chat.Q8_0/RoLlama2-7b-Chat.Q8_0-00001-of-XXXXX.gguf RoLlama2-7b-Chat.Q8_0.gguf
   • Make sure to point gguf-split to the first chunk of the split.

---

Got a suggestion? Ping me @legraphista!