RichardErkhov/Azazelle_-_Argetsu-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Argetsu - GGUF

• Model creator: https://huggingface.co/Azazelle/
• Original model: https://huggingface.co/Azazelle/Argetsu/

Name	Quant method	Size
Argetsu.Q2_K.gguf	Q2_K	2.53GB
Argetsu.IQ3_XS.gguf	IQ3_XS	2.81GB
Argetsu.IQ3_S.gguf	IQ3_S	2.96GB
Argetsu.Q3_K_S.gguf	Q3_K_S	2.95GB
Argetsu.IQ3_M.gguf	IQ3_M	3.06GB
Argetsu.Q3_K.gguf	Q3_K	3.28GB
Argetsu.Q3_K_M.gguf	Q3_K_M	3.28GB
Argetsu.Q3_K_L.gguf	Q3_K_L	3.56GB
Argetsu.IQ4_XS.gguf	IQ4_XS	3.67GB
Argetsu.Q4_0.gguf	Q4_0	3.83GB
Argetsu.IQ4_NL.gguf	IQ4_NL	3.87GB
Argetsu.Q4_K_S.gguf	Q4_K_S	3.86GB
Argetsu.Q4_K.gguf	Q4_K	4.07GB
Argetsu.Q4_K_M.gguf	Q4_K_M	4.07GB
Argetsu.Q4_1.gguf	Q4_1	4.24GB
Argetsu.Q5_0.gguf	Q5_0	4.65GB
Argetsu.Q5_K_S.gguf	Q5_K_S	4.65GB
Argetsu.Q5_K.gguf	Q5_K	4.78GB
Argetsu.Q5_K_M.gguf	Q5_K_M	4.78GB
Argetsu.Q5_1.gguf	Q5_1	5.07GB
Argetsu.Q6_K.gguf	Q6_K	5.53GB
Argetsu.Q8_0.gguf	Q8_0	7.17GB

Original model description:

pipeline_tag: text-generation tags: - mistral - merge license: cc-by-4.0

Model Card for Argetsu

Slerp merge of lots of models.

.yaml file for mergekit

slices:
  - sources:
      - model: Azazelle/SlimMelodicMaid
        layer_range: [0, 32]
      - model: Azazelle/Dumb-Maidlet
        layer_range: [0, 32]
merge_method: slerp
base_model: mistralai/Mistral-7B-v0.1
parameters:
  t:
    - filter: self_attn
      value: [0.2, 0.6, 0.44, 0.76, 1]
    - filter: mlp
      value: [0.8, 0.4, 0.56, 0.24, 0]
    - value: 0.5 # fallback for rest of tensors
dtype: float16