RichardErkhov/Sao10K_-_Euryale-1.4-L2-70B-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Euryale-1.4-L2-70B - GGUF

• Model creator: https://huggingface.co/Sao10K/
• Original model: https://huggingface.co/Sao10K/Euryale-1.4-L2-70B/

Name	Quant method	Size
Euryale-1.4-L2-70B.Q2_K.gguf	Q2_K	23.71GB
Euryale-1.4-L2-70B.IQ3_XS.gguf	IQ3_XS	26.37GB
Euryale-1.4-L2-70B.IQ3_S.gguf	IQ3_S	27.86GB
Euryale-1.4-L2-70B.Q3_K_S.gguf	Q3_K_S	27.86GB
Euryale-1.4-L2-70B.IQ3_M.gguf	IQ3_M	28.82GB
Euryale-1.4-L2-70B.Q3_K.gguf	Q3_K	30.99GB
Euryale-1.4-L2-70B.Q3_K_M.gguf	Q3_K_M	30.99GB
Euryale-1.4-L2-70B.Q3_K_L.gguf	Q3_K_L	33.67GB
Euryale-1.4-L2-70B.IQ4_XS.gguf	IQ4_XS	34.64GB
Euryale-1.4-L2-70B.Q4_0.gguf	Q4_0	36.2GB
Euryale-1.4-L2-70B.IQ4_NL.gguf	IQ4_NL	36.55GB
Euryale-1.4-L2-70B.Q4_K_S.gguf	Q4_K_S	36.55GB
Euryale-1.4-L2-70B.Q4_K.gguf	Q4_K	38.58GB
Euryale-1.4-L2-70B.Q4_K_M.gguf	Q4_K_M	38.58GB
Euryale-1.4-L2-70B.Q4_1.gguf	Q4_1	40.2GB
Euryale-1.4-L2-70B.Q5_0.gguf	Q5_0	44.2GB
Euryale-1.4-L2-70B.Q5_K_S.gguf	Q5_K_S	44.2GB
Euryale-1.4-L2-70B.Q5_K.gguf	Q5_K	45.41GB
Euryale-1.4-L2-70B.Q5_K_M.gguf	Q5_K_M	45.41GB
Euryale-1.4-L2-70B.Q5_1.gguf	Q5_1	48.2GB
Euryale-1.4-L2-70B.Q6_K.gguf	Q6_K	52.7GB
Euryale-1.4-L2-70B.Q8_0.gguf	Q8_0	68.26GB

Original model description:

license: llama2 language: - en

gguf quants: https://huggingface.co/Sao10K/Euryale-1.4-L2-70B-GGUF

1.3, but better? I guess.

Base Merged Model ratios adjusted.

NSFL portion of Hesperus v1 dataset trained and applied.

LimaRP merged in at a ~25% weight at the end.

Subjectively better in some aspects eg. long form rp, worse than the other, eg. chat-style rps.

overall a minor improvement in my eyes.

1.5 will include Hesperus v2 dataset in its entirety.

format: alpaca.