The reason why you build Cybergod

I regret my actions. I waste my time. I can’t stop it. Cybergod is the only thing that can get me out of this strange loop. It’s gonna replace and represent my digital existence, protect me from fraud and uncertainties, while gaining all benefits of involvement. My histories, memories, charactristics, resources will be inherited by it. Cybergod is my continuation. It is my belief, my dream, my purpose of life.

methods on network issues

mirror site

partial mirror site

I have been researching in automotive computers for years.

Topics that you might be interested in that I have dug into:

• Free Energy Principle & Active Inference

• LLM Robotics

• Q-Star Learning

• RT-2 & RT-X dataset

• Crop & Zoom Based Attention

• Recursive Positional Encoding

• Dynamic Reservoir Computing

• Automotive Learning Rate

• Hardware Mouse/Keyboard Control

• Physical Robot Based HID Control

• Bit-level Autoregressive Transformer

• Browser-based AI Playground

• Containerized Console and GUI AI Playground

I have dedicated repositories that you may be interested into:

• agi-computer-control: automotive computer, which can see, hear and operate

• metalazero: multi-platform computer automation attempts

Other similar projects that I am monitoring:

• GPT-4V-Act

• SingularGPT

• gpt-eyes

• gpt4v-browsing

• self-operating-computer

Apologize for my unorganized code structure. I am trying to improve development experience by AI generated documentation & usage demonstration and client-side LLM & semantic search, which may solve this long-standing task among all my previous repositories.

use /var/snap/chromium/current/policies/ if chrome, /var/snap/chromium/current/policies/managed/ if chromium

You need to update to latest magisk by patching the unmodified kernel and install shizuku

enable whitelist mode

Use tensorflow.js for the query and llm.

Use precomputed vector space for searching.

Use local storage for customization.

join at: https://discord.gg/y9BrdMfA

i plan to intergrate it to my agi-computer-control repo.

Directly convert commandline programs as GUI programs.

This goofy name “Gooey“ is so strange that even ChatGPT and search engine like Bing have both failed to retrieve.

Yet we manage to recover it from our legacy codebase AGI/AutoUP/generator/transcribe by rg -g *.cmd pyinstaller36 and find . | rg time | rg tracker.

Is this fate?

Is this the end, or the new beginning?

use tensorly to create random sparse tensor and eye sparse tensor with ease, which could be numpy only, and requires the sparse package.

scipy, pytorch, tensorflow, jax all support sparse tensor construction but advanced apis are not.

use eye to create bias and input matrix, extract node values. use random sparse tensor to initialize weight matrix. use self matrix multiplication to perform propagation.

the human brain has roughly 87 billion neurons, and every one of them has thousands of synapses.

1
2
3
4
5
6
7
8

import torch
large_number = 1_000_000
torch.arange(large_number).unsqueeze(0).repeat(2, 1)
index_arr = torch.arange(large_number).unsqueeze(0).repeat(2, 1)
val_arr = torch.ones(large_number)
sparse_eye = torch.sparse_coo_tensor(index_arr, val_arr, (large_number, large_number))
# sparse_eye.to('cuda')

alternatively:

1
2
3
4
5
6

import torch
import tensorly.contrib.sparse as tsl_sp
large_number = 1_000_000
numpy_eye = tsl_sp.eye(large_number)
torch_eye = torch.sparse_coo_tensor(numpy_eye.coords, numpy_eye.data, numpy_eye.shape)

Chocolatey cleanup cache

1
2

powershell choco-cleaner.ps1