Quantization made by Richard Erkhov.

ko-solar-10.7b-v0.8 - GGUF

Model creator: https://huggingface.co/mssma/
Original model: https://huggingface.co/mssma/ko-solar-10.7b-v0.8/

Name	Quant method	Size
ko-solar-10.7b-v0.8.Q2_K.gguf	Q2_K	3.79GB
ko-solar-10.7b-v0.8.IQ3_XS.gguf	IQ3_XS	4.21GB
ko-solar-10.7b-v0.8.IQ3_S.gguf	IQ3_S	4.44GB
ko-solar-10.7b-v0.8.Q3_K_S.gguf	Q3_K_S	4.41GB
ko-solar-10.7b-v0.8.IQ3_M.gguf	IQ3_M	4.58GB
ko-solar-10.7b-v0.8.Q3_K.gguf	Q3_K	4.91GB
ko-solar-10.7b-v0.8.Q3_K_M.gguf	Q3_K_M	4.91GB
ko-solar-10.7b-v0.8.Q3_K_L.gguf	Q3_K_L	5.33GB
ko-solar-10.7b-v0.8.IQ4_XS.gguf	IQ4_XS	5.5GB
ko-solar-10.7b-v0.8.Q4_0.gguf	Q4_0	5.73GB
ko-solar-10.7b-v0.8.IQ4_NL.gguf	IQ4_NL	5.8GB
ko-solar-10.7b-v0.8.Q4_K_S.gguf	Q4_K_S	5.78GB
ko-solar-10.7b-v0.8.Q4_K.gguf	Q4_K	6.1GB
ko-solar-10.7b-v0.8.Q4_K_M.gguf	Q4_K_M	6.1GB
ko-solar-10.7b-v0.8.Q4_1.gguf	Q4_1	6.35GB
ko-solar-10.7b-v0.8.Q5_0.gguf	Q5_0	6.97GB
ko-solar-10.7b-v0.8.Q5_K_S.gguf	Q5_K_S	6.97GB
ko-solar-10.7b-v0.8.Q5_K.gguf	Q5_K	7.16GB
ko-solar-10.7b-v0.8.Q5_K_M.gguf	Q5_K_M	7.16GB
ko-solar-10.7b-v0.8.Q5_1.gguf	Q5_1	7.59GB
ko-solar-10.7b-v0.8.Q6_K.gguf	Q6_K	8.29GB
ko-solar-10.7b-v0.8.Q8_0.gguf	Q8_0	10.74GB

Original model description:

library_name: transformers license: apache-2.0 language: - ko

usage


from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

path = "mssma/ko-solar-10.7b-v0.8"
model = AutoModelForCausalLM.from_pretrained(
        path,
        return_dict=True,
        torch_dtype=torch.float16,
        device_map='auto'
)
tokenizer = AutoTokenizer.from_pretrained(path)