V0.2 is the second training run with 27 images (x 5 repeats x 10 epochs) featuring only Ayrton Senna as driver (I'm sorry for Alain Prost but Ayrton is too iconic..) and with hand made captions and with cherry picked regularisation images, against SDXL 1.0 (VAE fix) base model.Apparently with the Prodigy optimizer seems important to always keep the LoRA weight at 1, I want to do a new training session with more steps to better understand this behaviour..In the previous version, identical in terms of parameters, but without regularisation images the wheels displacement was continuously a mess, now it seems that this has been improved.As always all my models are works in progress, with plenty of room for improvement. Every suggestion and feedback is very welcome.. this is my main reason that drives me to be part of this community.The McLaren MP4/4, also known as the McLaren-Honda MP4/4, is one of the most successful Formula One car designs of all time. Powered by Honda's RA168E 1.5-litre V6-turbo engine and driven by teammates Alain Prost and Ayrton Senna, the car competed during the 1988 Formula One season. The design of the car was led by American engineer Steve Nichols.At the end of the season, McLaren-Honda had taken both the Constructors and Drivers' titles (Senna edging out Prost due to winning more races - only the eleven best results counted, so even though Prost scored more overall points, Senna's 8 first-place finishes to Prost's 7, meant Senna took the championship title.) McLaren-Honda, who scored a then-record 199 points in the Constructors Championship, wrapped up the Constructors title with a 1-2 finish in Belgium for Round 11 of the 16 race season, it was the team's eighth 1-2 finish of the season (Senna and Prost would finish 1-2 twice more, in Japan and Australia). The team finished the season a massive 134 points in front of 2nd placed Ferrari.

Senna's McLaren MP4/4 Formula One (1988)

27 images (x 5 repeats) with hand made captions and cherry picked regularisation imagesKohya settings:<ul><li>Optimizer: Prodigy</li><li>LR/TE: 1</li><li>Scheduler: Cosine</li><li>Train batch size: 1</li></ul>

Files with SHA256: 796906b974e1b6e7c2374fb30adf7ad7418289f436af733046d6b5b3775bde28