metadata

base_model:
  - Bacon666/Phenom-12B-0.1
  - benhaotang/nemo-math-science-philosophy-12B
  - FallenMerick/MN-Chunky-Lotus-12B
  - GalrionSoftworks/Canidori-12B-v1
  - GalrionSoftworks/Pleiades-12B-v1
  - Luni/StarDust-12b-v2
  - Nohobby/InsanityB
  - Nohobby/MN-12B-Siskin-v0.2
  - ProdeusUnity/Stellar-Odyssey-12b-v0.0
  - Pyroserenus/Orthrus-12b-v0.8
  - rityak/MN-Maghin-12B
  - rityak/MN-RocinanteCelestar-12B
  - royallab/MN-LooseCannon-12B-v2
  - spow12/ChatWaifu_12B_v2.0
  - Svak/MN-12B-Inferor-v0.0
  - ThijsL202/MadMix-Unleashed-12B
  - Trappu/Abomination-merge-attempt-12B
  - VongolaChouko/Starcannon-Unleashed-12B-v1.0
library_name: transformers
tags:
  - mergekit
  - merge
  - bfloat16
  - safetensors
  - 12b
  - chat
  - creative
  - roleplay
  - conversational
  - creative-writing
  - not-for-all-audiences
language:
  - en
  - ru

DarkAtom-12B-v3

Something that shouldn't exist.

This is an interesting merge of 18 cool models, created using mergekit. It took quite a bit of my time, mostly due to the limitations of my old hardware, but I think it was definitely worth it. Enjoy exploring :)

Merge Details

Method

This model was merged using the multistep (Slerp|ModelStock|Ties) process and remerge with some model variations for best result.

Models

The following models were included in the merge:

Configuration

The following YAML configurations was used to produce this model. Some parameters may have diffirent pattern, but its not important to understand my workflow.

# Generation_1 from 18 original models:
models:
  - model: Original_Model_M
  - model: Original_Model_K
merge_method: slerp
base_model: Original_Model_M
dtype: bfloat16
parameters:
  t: [0.1, 0.9, 0.1, 0.9, 0.1, 0.9, 0.1, 0.9, 0.1, 0.9, 0.1, 0.9]

# Variant_N from Generation_1 and AlphaMerge:
models:
  - model: SecretModel_A
    parameters:
      density: [0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1]
      weight:  [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]
  - model: SecretModel_B
    parameters:
      density: [0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2]
      weight:  [0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8]
  - model: SecretModel_C
    parameters:
      density: [0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3]
      weight:  [0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7]
  - model: SecretModel_D
    parameters:
      density: [0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4]
      weight:  [0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6]
  - model: SecretModel_E
    parameters:
      density: [0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5]
      weight:  [0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5]
  - model: SecretModel_F
    parameters:
      density: [0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6]
      weight:  [0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4]
  - model: SecretModel_G
    parameters:
      density: [0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7]
      weight:  [0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3]
  - model: SecretModel_H
    parameters:
      density: [0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8]
      weight:  [0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2]
  - model: SecretModel_I
    parameters:
      density: [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]
      weight:  [0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1]
merge_method: ties
base_model: AlphaMerge
dtype: bfloat16

# Model stock merge for create:
# + Generation_2 from SecretModels
# + Variant_M from Generation_2
# + AlphaMerge from intuitively selected and forgotten models
models:
  - model: SecretModel_A
  - model: SecretModel_B
  - model: SecretModel_C
merge_method: model_stock
base_model: SecretModel_A
dtype: bfloat16

# Final Variant from Variant_N, Variant_M, and one good model from Generation_1:
models:
  - model: Variant_N
    parameters:
      density: [0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1]
      weight:  [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]
  - model: Good_G1_Model
    parameters:
      density: [0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2]
      weight:  [0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.8]
merge_method: ties
base_model: Variant_M
dtype: bfloat16

My thanks to the authors of the original models, your work is incredible. Have a good time 🖤