Neural Forge - AI Training Idle Game

AI Core

Lv. 0

Basic Model

Training Progress 0%

Tokens/sec

0

Intel/sec

0

Research Tree

Architecture

Multi-Layer Perceptron

Basic neural network architecture

Unlocked

Convolutional Network

Better for pattern recognition

Cost: 100 intel

Training Techniques

Stochastic Gradient Descent

Basic optimization algorithm

Unlocked

Adam Optimizer

More efficient training

Cost: 150 intel

Hardware

GPU Acceleration

2x training speed

Cost: 200 intel

TPU Cluster

5x training speed

Locked

Data Streams

MNIST Dataset

Active

Handwritten digit recognition

+1 token/sec 0.5 intel/sec

CIFAR-10

Locked

Object recognition in images

+5 token/sec 2 intel/sec

Wikipedia Corpus

Locked

Natural language processing

+20 token/sec 10 intel/sec

Model Output

[System] Initializing neural network...

[System] Loading training data...

AI Core

Research Tree

Multi-Layer Perceptron

Convolutional Network

Stochastic Gradient Descent

Adam Optimizer

GPU Acceleration

TPU Cluster

Data Streams

MNIST Dataset

CIFAR-10

Wikipedia Corpus

Model Output

Milestones