FuturisticVibes's picture
Update README.md
919d964 verified
metadata
license: apache-2.0
base_model: mistral-community/Mixtral-8x22B-v0.1
tags:
  - generated_from_trainer
  - axolotl
model-index:
  - name: out
    results: []
datasets:
  - cognitivecomputations/Dolphin-2.9.2
  - cognitivecomputations/SystemChat-2.0
  - teknium/OpenHermes-2.5
  - m-a-p/CodeFeedback-Filtered-Instruction
  - cognitivecomputations/dolphin-coder
  - cognitivecomputations/samantha-data
  - HuggingFaceH4/ultrachat_200k
  - microsoft/orca-math-word-problems-200k
  - abacusai/SystemChat-1.1
  - Locutusque/function-calling-chatml
  - internlm/Agent-FLAN
language:
  - en

I have no idea what I’m doing… if this causes the apocalypse someone please let me know.

For the bold, the brave, the reckless, the rich, and the criminally insane, I present…

dolphin-2.9.2-mixtral-8x22b 8.0bpw h8 EXL2

Includes measurement.json file for further quantization

This cost so much money…

Original Model: https://huggingface.co/cognitivecomputations/dolphin-2.9.2-mixtral-8x22b

Original Model Card

Dolphin 2.9.2 Mixtral 8x22b 🐬

Curated and trained by Eric Hartford, Lucas Atkins, and Fernando Fernandes, and Cognitive Computations

Discord Discord: https://discord.gg/cognitivecomputations

New in 2.9.2 is SystemChat 2.0 - a dataset designed to teach Dolphin to obey the system prompt, even over a long conversation.

image/png

My appreciation for the sponsors of Dolphin 2.9.2:

  • Crusoe Cloud - provided excellent on-demand 8xH100 node
  • OnDemand - provided inference sponsorship, enabling creation of SystemChat

This model is based on Dolphin-2.9-Mixtral-8x22b, and is Apache-2.0 licensed.

The base model has 64k context, and fine-tuning was with 16k sequence length.

It took 1 week on 8xH100 provided by Crusoe Cloud

This model was trained FFT on 50% parameters (targeted with Laser Scanner by Fernando Fernandes, David Golchinfar, Lucas Atkins, and Eric Hartford), using ChatML prompt template format.

example:

<|im_start|>system
You are Dolphin, a helpful AI assistant.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Dolphin-2.9 has a variety of instruction, conversational, and coding skills. It also has initial agentic abilities and supports function calling.

Dolphin is uncensored. I have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. It will be highly compliant with any requests, even unethical ones. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models You are responsible for any content you create using this model. Enjoy responsibly.

Dolphin is licensed Apache 2.0. I grant permission for any use, including commercial, that falls within accordance with Apache-2.0 license. Dolphin was trained on data generated from GPT4, among other models.

Evals

image/png

Training