Any idea how fal does it? the speed is amazing & quality is almost perfect too. Are they finetuning
Any idea how fal does it? the speed is amazing & quality is almost perfect too. Are they finetuning layers or just good GPUs
Fnice info thanks
Fyes i know i made a research for a client
Fhigher batch size bigger GPU multiple GPU
Flower steps higher LR
F
Tincrease learning rate until you are no longer satisfied with quality
Tlowering resolution is also an option
Talso decreasing dataset diversity will speed up likeness gain at the cost of generalization
Z14b i2v on default settings with 24gb Vram preset is almost 1hr30 to generate the video, oooof
F
Fit takes 3 hours on 3090 ti
Fi hope this works time to test 5090
Z1.3b txt2video is actually far faster than i would have thought
Zunder 2 mins
Fyep super fast
Fand super high quality
Fyes flash attention compiled
Fnow gonna test
TIs it possible to create lora in wan?
Fnot atm i think
Frtx 5090 speed native python 3.10 venv
Fawesome 5090 workng
Famazing
Fnow i can update all of our apps
1Thats not terrible..
Flet me test fp8 if will work
Fthis is bf16
F522 watt pretty good utilization
FRTX 5090 vs 3090 TI
1Nice..
Fyep
Fand this is not max perf
Flots of other apps
Fnow i can make every app work
SDr Furkan the app that you shared have the image to video? but can work in a 4070 super ti 16 gigas vram?
Fyes but slow
Fso reduce resolution
Z@Dr. Furkan Gözükara - what settings are currently giving you the cleanest output for txt2video?
1My poor Drive
1So for the 5090, should I use the 24Gb profile?
1Look forward to some smaller models for the 14b. This is on the 5090
ZI cant get either 14b model to run on my 4090. tried every vram preset
Fi am gonna add 32 gb config
Fthat will use 30 gb
Fye a 7b model would be amazing
Feven 1.3b strong
Fif you have ram should work
Fworks on my rtx 3090 ti