133 lines
5.6 KiB
Markdown
133 lines
5.6 KiB
Markdown
---
|
|
base_model:
|
|
- LeroyDyer/Mixtral_AI_Multi_TEST
|
|
- LeroyDyer/Mixtral_AI_Cyber_Dolphin_2.0
|
|
- LeroyDyer/Mixtral_AI_CyberLAW
|
|
- LeroyDyer/Mixtral_AI_CyberBrain_3_0
|
|
- LeroyDyer/Mixtral_AI_Cyber_5.0
|
|
- LeroyDyer/Mixtral_AI_CyberBrain_2.0
|
|
- ezelikman/quietstar-8-ahead
|
|
- KoboldAI/Mistral-7B-Erebus-v3
|
|
library_name: transformers
|
|
tags:
|
|
- mergekit
|
|
- megamerge
|
|
- code
|
|
- Cyber-Series
|
|
license: mit
|
|
language:
|
|
- en
|
|
datasets:
|
|
- Open-Orca/OpenOrca
|
|
- cognitivecomputations/dolphin
|
|
- WhiteRabbitNeo/WRN-Chapter-2
|
|
- WhiteRabbitNeo/WRN-Chapter-1
|
|
- gate369/Alpaca-Star
|
|
- gate369/alpaca-star-ascii
|
|
---
|
|
|
|
<img src="https://cdn-avatars.huggingface.co/v1/production/uploads/65d883893a52cd9bcd8ab7cf/tRsCJlHNZo1D02kBTmfy9.jpeg" width="300"/>
|
|
https://github.com/spydaz
|
|
|
|
Currently undegoing Fine tuning ! as this model contains all Previous models !
|
|
|
|
|
|
This model contains many hidden tensors :
|
|
As it was emrged with many lora adapter for various task such as vision and sound .
|
|
The problem was that for some reason i could not get the extra heads to show up like other models.
|
|
such as the llava model ... i suppose this model can change the config.json to be a llava model and yes ! it works! ie it can think and has hidden think heads ? but you need to config it up !, It has vision heads but also i could not set the config up !
|
|
so hidden talents:
|
|
It was also merged with the mothers of these models for QUiet(thoughts) and (llava vision etc ) so the tensors are there . i just did not understand how to fine tne the addtional funcitonalitys. as they need a single trainign example to populate the hidden tensor hence te merges. and yet when the model is put in train mode , ie by setting the model after loading to model.TRAIN ... the tensors apear waiting for training so just add a peft and start the training!
|
|
|
|
|
|
THIS VERSION HAS BEEN UPDATED TO INCLUDE CYBERBRAIN ! (Hidden Tensors)
|
|
|
|
## Extended capabilities:
|
|
* mistralai/Mistral-7B-Instruct-v0.1 - Prime-Base
|
|
|
|
* ChaoticNeutrals/Eris-LelantaclesV2-7b - role play
|
|
|
|
* ChaoticNeutrals/Eris_PrimeV3-Vision-7B - vision
|
|
|
|
* rvv-karma/BASH-Coder-Mistral-7B - coding
|
|
|
|
* Locutusque/Hercules-3.1-Mistral-7B - Unhinging
|
|
|
|
* KoboldAI/Mistral-7B-Erebus-v3 - NSFW
|
|
|
|
* Locutusque/Hyperion-2.1-Mistral-7B - CHAT
|
|
|
|
* Severian/Nexus-IKM-Mistral-7B-Pytorch - Thinking
|
|
|
|
* NousResearch/Hermes-2-Pro-Mistral-7B - Generalizing
|
|
|
|
* mistralai/Mistral-7B-Instruct-v0.2 - BASE
|
|
|
|
* Nitral-AI/ProdigyXBioMistral_7B - medical
|
|
|
|
* Nitral-AI/Infinite-Mika-7b - 128k - Context Expansion enforcement
|
|
|
|
* Nous-Yarn-Mistral-7b-128k - 128k - Context Expansion
|
|
|
|
* yanismiraoui/Yarn-Mistral-7b-128k-sharded
|
|
|
|
* ChaoticNeutrals/Eris_Prime-V2-7B - Roleplay
|
|
|
|
|
|
This Expert is a companon to the MEGA_MIND 24b CyberSeries represents a groundbreaking leap in the realm of language models, integrating a diverse array of expert models into a unified framework. At its core lies the Mistral-7B-Instruct-v0.2, a refined instructional model designed for versatility and efficiency.
|
|
|
|
Enhanced with an expanded context window and advanced routing mechanisms, the Mistral-7B-Instruct-v0.2 exemplifies the power of Mixture of Experts, allowing seamless integration of specialized sub-models. This architecture facilitates unparalleled performance and scalability, enabling the CyberSeries to tackle a myriad of tasks with unparalleled speed and accuracy.
|
|
|
|
Among its illustrious sub-models, the OpenOrca - Mistral-7B-8k shines as a testament to fine-tuning excellence, boasting top-ranking performance in its class. Meanwhile, the Hermes 2 Pro introduces cutting-edge capabilities such as Function Calling and JSON Mode, catering to diverse application needs.
|
|
|
|
Driven by Reinforcement Learning from AI Feedback, the Starling-LM-7B-beta demonstrates remarkable adaptability and optimization, while the Phi-1.5 Transformer model stands as a beacon of excellence across various domains, from common sense reasoning to medical inference.
|
|
|
|
With models like BioMistral tailored specifically for medical applications and Nous-Yarn-Mistral-7b-128k excelling in handling long-context data, the MEGA_MIND 24b CyberSeries emerges as a transformative force in the landscape of language understanding and artificial intelligence.
|
|
|
|
Experience the future of language models with the MEGA_MIND 24b CyberSeries, where innovation meets performance, and possibilities are limitless.
|
|
### Models Merged
|
|
|
|
The following models were included in the merge:
|
|
* [LeroyDyer/Mixtral_AI_Multi_TEST](https://huggingface.co/LeroyDyer/Mixtral_AI_Multi_TEST)
|
|
* [LeroyDyer/Mixtral_AI_CyberLAW](https://huggingface.co/LeroyDyer/Mixtral_AI_CyberLAW)
|
|
* [LeroyDyer/Mixtral_AI_CyberBrain_3_0](https://huggingface.co/LeroyDyer/Mixtral_AI_CyberBrain_3_0)
|
|
* [LeroyDyer/Mixtral_AI_Cyber_5.0](https://huggingface.co/LeroyDyer/Mixtral_AI_Cyber_5.0)
|
|
|
|
### Configuration
|
|
|
|
The following YAML configuration was used to produce this model:
|
|
|
|
```yaml
|
|
|
|
models:
|
|
- model: LeroyDyer/Mixtral_AI_Cyber_Dolphin_2.0
|
|
parameters:
|
|
density: [0.256, 0.512, 0.128] # density gradient
|
|
weight: 0.382
|
|
- model: LeroyDyer/Mixtral_AI_CyberLAW
|
|
parameters:
|
|
density: 0.382
|
|
weight: [0.256, 0.128, 0.256, 0.128] # weight gradient
|
|
- model: LeroyDyer/Mixtral_AI_CyberBrain_3_0
|
|
parameters:
|
|
density: 0.382
|
|
weight: [0.128, 0.512, 0.128, 0.128] # weight gradient
|
|
- model: LeroyDyer/Mixtral_AI_Multi_TEST
|
|
parameters:
|
|
density: 0.382
|
|
weight: [0.128, 0.512, 0.128, 0.128] # weight gradient
|
|
- model: LeroyDyer/Mixtral_AI_Cyber_5.0
|
|
parameters:
|
|
density: 0.382
|
|
weight:
|
|
- filter: mlp
|
|
value: 0.5
|
|
- value: 0
|
|
merge_method: ties
|
|
base_model: LeroyDyer/Mixtral_AI_Cyber_Dolphin_2.0
|
|
parameters:
|
|
normalize: true
|
|
int8_mask: true
|
|
dtype: float16
|
|
|
|
``` |