Krizhevsky spent the next year constantly tweaking parameters and retraining the network

Evidence

CHM Releases AlexNet Source Code (Computer History Museum)

Categorization

• corroborated: 2 independent passages support this claim
• XREF: connects to: Krizhevsky trained an early GPU-based neural network for two weeks in 2012

Related

• Krizhevsky trained an early GPU-based neural network for two weeks in 2012 (similarity 0.91)
• Krizhevsky trained an early GPU-based deep neural network in his bedroom in 2012 (similarity 0.86)
• AlexNet was trained on two NVIDIA GTX 580 GPUs for several weeks (similarity 0.86)