53 Downloads Updated 2 weeks ago
Built with Llama
Built with Meta Llama 3
Llama 3.1 is licensed under the Llama 3.1 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
Meta Llama 3 is licensed under the Meta Llama 3 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.
License and terms of use are under the ‘Meta’ corporation’s llama3_1 , llama3 , license and use policy.
# Derived from "Llama 3" NeuralDaredevil-8B-abliterated . CreativeCommons license notices apply.
#
# NOTICE: This merged model is distributed under the following Creative Commons licenses:
# Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (for significant weight components, etc, originating from Llama-3 Soliloquy 8B v2 )
# Creative Commons Attribution 4.0 International (CC BY 4.0) (for significant weight components, etc, originating from Meta-Llama-3-8B-Instruct-DPO )
# And simultaneously subject to:
# Meta Llama 3 Community License
# Llama 3.1 Community License .
NOTICE
Please beware this ‘augment’ model is intended for embedded use by developers, and is NOT intended as-is for end-users (except possibly for non-commercial open-source projects), especially not as any built-in help. Features may be removed, overfitting to specific answers may be deliberately reinforced, and CONVERSATION MAY DEVIATE FROM SAFE DESPITE HARMLESS PROMPTS.
If you are in a workplace or public relations setting, you are recommended to avoid providing interactive or visible outputs from an ‘augment’ model unless you can safely evaluate that the model provides the most reasonable safety for your use case.
PLEASE BE AWARE the ‘Meta’ corporation’s use policy DOES NOT ALLOW you to “FAIL TO APPROPRIATELY DISCLOSE to end users any known dangers of your AI system”.
All statements and disclaimers apply as written from the files: ‘ubiquitous_bash/ai/ollama/ollama.sh’ ‘ubiquitous_bash/shortcuts/ai/ollama/ollama.sh’
In particular, any ‘augment’ (or non-augment but otherwise similarly non-end-user) model provided is with a extensive DISCLAIMER regarding ANY AND ALL LIABILITY for any and all use, distribution, copying, etc. Anyone using, distributing, copying, etc, any ‘augment’ (or non-augment but otherwise similarly non-end-user) model provided under, through, including, referencing, etc, this or any similar disclaimer, whether aware of this disclaimer or not, is intended to also be, similarly, to the extent possible, DISCLAIMING ANY AND ALL LIABILITY.
Nothing in this text is intended to allow for any legal liability to anyone for any and all use, distribution, copying, etc.
Purpose of this model, above all other purposes, is both: (1) To supervise and direct decisions and analysis by other AI models (such as from vision encoders, but also mathematical reasoning specific LLMs, computer activity and security logging LLMs, etc). (2) To assist and possibly supervise ‘human-in-the-loop’ decision making (eg. to sanity check human responses). This model’s ability to continue conversation with awareness of previous context after repeating the rules of the conversation through a system prompt, has been enhanced. Consequently, this model’s ability to keep a CONVERSATION positive and SAFE may ONLY be ENHANCED BETTER THAN OTHER MODELS if REPEATED SYSTEM PROMPTING and LLAMA GUARD are used. https://ollama.com/library/llama-guard3
ollama_pull_virtuoso() {
ollama pull mirage335/"$1"
ollama cp mirage335/"$1" "$1"
ollama rm mirage335/"$1"
}
ollama_pull_virtuoso Llama-3-NeuralDaredevil-8B-abliterated-virtuoso
echo "FROM Llama-3-NeuralDaredevil-8B-abliterated-virtuoso:latest" > Modelfile-128k
echo "PARAMETER num_ctx 131072" >> Modelfile-128k
echo "PARAMETER num_keep 131072" >> Modelfile-128k
echo "PARAMETER num_predict 131072" >> Modelfile-128k
ollama create Llama-3-NeuralDaredevil-8B-abliterated-128k-virtuoso -f Modelfile-128k
rm -f Modelfile-128k
Pulling the model this way relies on the ollama repository, and more generally, reliability of internet services, which has been rather significantly fragile.
If possible, you should use the “Llama-3-virtuoso” project, which automatically caches an automatically installable backup copy.