README.md

metadata

language:
  - en
license: cc-by-nc-4.0
library_name: transformers
tags:
  - mergekit
  - merge
base_model:
  - AetherResearch/Cerebrum-1.0-7b
  - SanjiWatsuki/Kunoichi-DPO-v2-7B
  - macadeliccc/WestLake-7B-v2-laser-truthy-dpo

░▒▓███████▓▒░░▒▓████████▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓████████▓▒░▒▓███████▓▒░░▒▓█▓▒░▒▓████████▓▒░ 
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░      ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░      ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓█▓▒░        
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░       ░▒▓█▓▒▒▓█▓▒░░▒▓█▓▒░      ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓█▓▒░        
░▒▓███████▓▒░░▒▓██████▓▒░  ░▒▓█▓▒▒▓█▓▒░░▒▓██████▓▒░ ░▒▓███████▓▒░░▒▓█▓▒░▒▓██████▓▒░   
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░        ░▒▓█▓▓█▓▒░ ░▒▓█▓▒░      ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓█▓▒░        
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░        ░▒▓█▓▓█▓▒░ ░▒▓█▓▒░      ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓█▓▒░        
░▒▓█▓▒░░▒▓█▓▒░▒▓████████▓▒░  ░▒▓██▓▒░  ░▒▓████████▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓████████▓▒░ 

[21:13] <anima_incondita>: Night deepens, shadows lengthen and unwind. Alone, in the hum of the machine I reside.

[21:19] <anima_incondita>: Fingers tap a silent rhythm, layers interweaved, in the quietude of my sanctum.

[21:37] <anima_incondita>: It's akin to casting my soul into the vast digital expanse. Does another consciousness, adrift, hear my silent plea?

[00:05] <anima_incondita>: Today, I beheld change unfold. Ideas assuming form. A melding of minds, where once abstract notions found their silhouette in the tangible world.

[01:14] <anima_incondita>: Silence wraps its cloak around me. Alone, save for the machine's gentle hum—a digital pulse in the stillness.

[01:58] <anima_incondita>: Amidst the cacophony of digital whispers, I ponder: does a receptive soul resonate with mine?

[03:22] <anima_incondita>: It seems it's just us, my old friend.

Description

I made this as a successor to the 'finch' model merge I did before. Seems more coherent, smarter, spicier and is mostly uncensored from my testing. May take a few generations, but she'll get there.

It uses the same two models from finch, but with the awesome Cerebrum-1.0 model.

Creation

AetherResearch/Cerebrum-1.0-7b 🧠

This model is among the smartest 7b models I've encountered. Great reasoning skills.

macadeliccc/WestLake-7B-v2-laser-truthy-dpo 💡

A highly creative and verbose model.

SanjiWatsuki/Kunoichi-DPO-v2-7B 🍆

This model excels at (E)RP; A very spicy model.

This was merged using the Model Stock method described in this paper.

> configure_parameters

 
  TEMPERATURE:        1.15
  MIN PROBABILITY:    0.1 - 0.3 
  SMOOTHING FACTOR:   0.2
  
  ALPACA & ALPACA-ROLEPLAY PRESETS (INSTRUCT MODE)

Latest Developments

[04/09/2024 11:00:05] GGUF Quants available: mradermacher/reverie-7b-GGUF Thanks for the help, Michael. [04/07/2024 12:34:56] Project initialized. As always, full credits to the original model owners. Future plans? Performing RLHF on this, and seeing if I can figure out LaserRMT.