kromeurus's picture
Update README.md
fb2b2df verified
|
raw
history blame
1.3 kB
metadata
base_model:
  - ArliAI/ArliAI-Llama-3-8B-Formax-v1.0
  - gradientai/Llama-3-8B-Instruct-Gradient-1048k
  - ArliAI/Llama-3.1-8B-ArliAI-Formax-v1.0
  - Sao10K/L3-8B-Niitama-v1
  - Sao10K/L3-8B-Stheno-v3.3-32K
  - tohur/natsumura-storytelling-rp-1.0-llama-3.1-8b
  - Sao10K/L3-8B-Tamamo-v1
  - vicgalle/Roleplay-Hermes-3-Llama-3.1-8B
library_name: transformers
tags:
  - mergekit
  - merge

Attempt number three (five) that aims to fix the overly chatty and flower language of v0.2.

Quants

OG Q8 GGUF by me.

Details & Recommended Settings

(Still testing; details subject to change)

Rec. Settings:

Template: L3
Temperature: 1.3
Min P: 0.1
Repeat Penalty: 1.05
Repeat Penalty Tokens: 256
Dyn Temp: 0.9-1.05 at 0.1
Smooth Sampl: 0.18

Merge Theory

Can't be asked rn.

base_model: merge/siithamol3.1
dtype: float32
merge_method: breadcrumbs
out_dtype: bfloat16
parameters:
  int8_mask: 1.0
  normalize: 0.0
slices:
- sources:
  - layer_range: [0, 32]
    model: merge/siithamol3.1
    parameters:
      density: 0.9
      gamma: 0.01
      weight: [0.5, 0.8, 0.9, 1.0]
  - layer_range: [0, 32]
    model: merge/formaxext.3.1
    parameters:
      density: 0.9
      gamma: 0.01
      weight: [0.5, 0.2, 0.1, 0.0]