Topics created by phillipsback

There something interesting.

DepthAnything can be done with THIS

In OpenAI API, I think Mr. uzuki_aoba already post in Qiita. In the CMD that renew the model list got some TYPO "sudo systemctrl restart llm-openai-api"

I try LLM(qwen3), Whisper(CN), MeloTTS(CN), VLM(InternVL), they are work great with StackFlow that I can call by another computer in my house. But I think it something funny when I install llm-model-internVL. In apt I called is internvl____axcl, but after install the folder name came to internvl____AX650. (In the end, I change the folder name manually...)

@phillipsback

Slider Odd (Auto jump to max. value. I think might be something wrong in update loop.) Can't success in using grad_dsc_t (can do in style way, but couldn't do in drawing API)

other's going well.
I prefer LVGL with micropython, althought slow in speed, and troublesome in callback function. But more Object Oriental to make coding logically (I think more clearly.)

And ...
It's so easy use LVGL in Tab5. Same version UIFlow2.0 firmware in Core2 AWE, have to run M5.begin first. (In someway , Tab5 might wrong in someway....)

@hacxx Dear hacxx, Thanks for your help.

UIFlow2 can feel odd due to inconsistent block behavior, limited error feedback, and occasional cloud sync issues, making it tricky for users to debug and manage projects efficiently.