Lukasz Czarnecki

Regular pedestrian

[Lukasz Czarnecki @lczarne]

@aaditsh Seeing many such posts triggering people - How do we know it's true?

[Lukasz Czarnecki @lczarne]

@SavannahFeder If a goal of such videos is to trigger some emotions and intentions and all such videos will be everywhere - how will it influence the process? Maybe making something outstanding and impressive will still take another scale of effort and budget?

[Lukasz Czarnecki @lczarne]

@JinaAI_ Can you describe from practice how well it can encode visual information? What depth of details is reasonable to assume? For example in more technical images like floor plans?

[Lukasz Czarnecki @lczarne]

@victormustar Looks promising. Is it possible to connect the process with own RAG setup?

[Lukasz Czarnecki @lczarne]

@kimmonismus I had an interesting discussion with o3 about open deep topics. After some exchanges suggested that there are limits in language capacity to cover how world works. There is clear truth in that. It's easy to see it comparing how different languages are able to cover the same info.

[Lukasz Czarnecki @lczarne]

@Hesamation @HeyNina101 Very cool. On viable suggestion I would have is to open it more to different LLM platformts so instead of particular clients files add a routhing file with platform parameters and function calls. In the end one would like to test/route 10+ different platforms.

[Lukasz Czarnecki @lczarne]

@BugNinza Is is possible to connect it with some RAG inside Jan app?

[Lukasz Czarnecki @lczarne]

@devloperVivek Cool any other tricks like that? Of using cursor internally not only to write code/text?

[Lukasz Czarnecki @lczarne]

@RahulKrishnaa28 Pretty cool trick! Any other like that with internal visualizations?

[Lukasz Czarnecki @lczarne]

@emilylambert @v_computer How are LLMs costs? This is my main concern with current AI Agents use cases. What was token usage in this case? Sustainable?

[Lukasz Czarnecki @lczarne]

@jasonzhou1993 What would be a real use case for using Suna? Where would it make sense practically?

[Lukasz Czarnecki @lczarne]

@jerryjliu0 Looks promising! Any clear advantages in comparison to using Unstract (paid tool)?

[Lukasz Czarnecki @lczarne]

@1romanprodan Basic RAG solution will not work for most of professional commercial cases. Reliance on comparing information based on embeddings will not work in documents dense in information and meaning. Graph RAG seems to be a direction. Problem still to be solved.

[Lukasz Czarnecki @lczarne]

@safwaankay Beam

[Lukasz Czarnecki @lczarne]

@EHuanglu Or this:

[Lukasz Czarnecki @lczarne]

@EHuanglu

[Lukasz Czarnecki @lczarne]

Internet: "Google’s Gemini 2.0: AI Image Generation & Editing is INSANE!"
Google Gemini 2.0:

[Lukasz Czarnecki @lczarne]

What will happen with your job in 2025-2030:
At least according to the World Economic Forum - Future of Jobs Report 2025.

[Lukasz Czarnecki @lczarne]

The new Chat GPT 4.5 is:
750 x more expensive than the Google Gemini Flash 2.0 model.
277 x more expensive than DeepSeek V3.
30 x more expensive than ChatGPT 4o.
25 x more expensive than ChatGPT Claude Sonnet 3.7.

In many aspects, these models are comparable.

[Lukasz Czarnecki @lczarne]

Bill Gates, przeszedł ostatnio glow up i zaczął podchodzić z większą fantazją do ubioru. To oczywiście nieprawda, tak jak coraz więcej treści wideo.
Nowe Kling AI Virtual Try On: daj mi osobę i zdjęcie ubioru, a pokażę Ci, jak ta prezentuje się w tej kreacji.