Alva Seltmann

physician, researcher. Currently diversity medicine @ruhrunibochum. Formerly AI and microscopy @AgEggeling

[Pablo Carravilla @carravilla_p]

It was great to contribute! Congratulations @aseltmannio and @dwaithe!

[Alva Seltmann @aseltmannio]

@IvanMasl0v @e_sisamakis @AgEggeling @carravilla_p @dwaithe @BiophysJ If you try the Colab and are running into problems - please reach out on Github and post an issue there, I will try to look into it :)

[Alva Seltmann @aseltmannio]

@IvanMasl0v @e_sisamakis @AgEggeling @carravilla_p @dwaithe @BiophysJ Great to See that you found the colab - yes, the underlying functions support batch import (you can find them in the Colab cells). I only implemented single file import in the Colab "User interface" because implementing the whole pipeline with a nice UI was a bit difficult

[Ivan Maslov @IvanMasl0v]

FCS-lovers struggling with challenging samples, check this:

"Neural network informed photon filtering reduces fluorescence correlation spectroscopy artifacts" by @aseltmannio @AgEggeling @carravilla_p @dwaithe

https://t.co/YnPAG5p40x @BiophysJ

GitHub: https://t.co/6B5rrPJsPF👏

[Dr Dominic Waithe @dwaithe]

Big congratulations to @aseltmannio for getting our paper over the finish line. @BiophysJ Neural network informed photon filtering reduces fluorescence correlation spectroscopy artifacts:
https://t.co/fq8gd9tuZF
@carravilla_p Katharina Reglinski @AgEggeling

[Alva Seltmann @aseltmannio]

Thanks for coming to our TED-Talk! 🦕

I want to attribute @skdh and @colin_fraser whose videos and texts helped me to wrap my head around these models.

25/25

[Alva Seltmann @aseltmannio]

In this year's group retreat @AgEggeling, Bela Vogler, @jacopo_abramo, and I prepared an Intro to #ChatGPT.

A 🧵on where we think it is useful, where dangerous, and which metaphors we liked for Large Language Models (#LLMS) and LLM-based tools like #Bard or #BingAI

1/n

[Alva Seltmann @aseltmannio]

My clear recommendation for scientists using these tools (with the above caveats in mind): make this transparent! At some point biases of these models will be hopefully known and texts which were written with their help can then be checked for them.

24/n