I was right again

Resumen

The creator explores the viral claim that feeding images instead of text to LLMs significantly improves efficiency and lowers costs. After performing extensive testing on his own tower defense game, he compares text-based context against image-based context. While the theory suggests massive performance gains, his results were mixed: images resulted in lower token usage but significantly slower processing times. He concludes that while the technology is promising for the future, current tools aren't fully optimized for this workflow, and developers should verify these claims with their own benchmarks rather than blindly following social media hype.

Capítulos

1. 0:00La superioridad de las imágenes
2. 0:43Análisis del paper de DeepSeek
3. 2:25Sponsor y optimización de datos
4. 5:11Teoría de la compresión óptica
5. 10:48Experimento con un juego

Temas tratados

• Efficiency of images vs text
• DeepSeek research on token density

+4 temas más (contenido completo al seguir el canal)

Extracto de la transcripción

You see that face? You know what that face is? That that is the smug face of somebody who was proven correct. And the best part, it's right after caught in 8K. You're even told by your friends you're wrong. So, you're probably thinking, "Okay, what what are we talking about? What are you even trying to say? What were you proven right about?" Well, my friends, it turns out after many hurt feelings and being told I was wrong, images might actually be better for models than text. I knew it. Don't make it do image recognition on >> No, no, you just do this. What are you that, dude. talking about? >> Hey, >> now this isn't just me saying this, of course. This is actually DeepSeek releasing a new paper that suggests that if you provide images, you could pull out 10 text token from a single image token with near 100% accuracy. In other words, a model's internal representation of an image is 10…