Frühere Suchanfragen
Suchoptionen
Verwaltet von:
Anoxinon e.V. 
If Codeberg is trying to "compete" against GitHub and GitLab, why does it refuse to take a look at AI assistants? Apart from infringing on authors' rights and questionable output quality, we think that the current hype wave led by major companies will leave a climate disaster in its wake: https://disconnect.blog/generative-ai-is-a-climate-disaster/
Other _sustainable_ (and cheaper!) ways for increasing efficiency in software development exist: In-project communication, powerful automation pipelines and reducing boilerplate.
As we have received a bunch of questions on the matter: If you'd like to explore machine learning and AI in the context of free and libre software (perhaps from a less commercialized perspective), you're absolutely welcome to use Codeberg.
We explicitly welcome researchers and solutions that are built more ethically. In general, we also love software that doesn't take lots of resources to run! As nuance can get lost fast, feel free to just ask us if you have any questions or concerns!
| 
| 
| 
@Codeberg@social.anoxinon.de
"AI in the context of free and libre software" = ∅
@Orca Check out this concept that uses LLMs to defend against AI scrapers, I am pretty sure you'd like it even if it falls under the category of "AI": https://codeberg.org/konterfai/konterfai
^n
@Codeberg@social.anoxinon.de
The core problem is, imho, most (if not all) LLM models are not possible to be made free/open-source. FOSS requires all parties being able to learn and develop with the knowledge shared in the FOSS source code, but afaik all these LLM models are "transparent" blobs (i.e. "blackboxes"), which made them impossible to be deciphered, and no one really knows how it worked (even those who trained the model), so only satisfying "the freedom to run the program as wish" in FSF's "4 essential freedoms of free software" (I do NOT endorse FSF but this concept is good by itself imo). And even so, there are models that are released under non-free licenses, like llama. llama's license is kind of similar to "Redis Source-Available License" imo, if model blobs can ever be counted as "source".
The project you mentioned, I saw that it depends on ollama (free software), which depends on LLM models like llama or phi-3. (Phi-3 is marked as MIT on Huggingface, though I doubt if anyone could benefit from its "programming") It's a neat project and has real use cases, but for me it's a bit like (no offense) installing Windows 11 on a computer with coreboot installed. Yes the loader (konterfai & ollama) is FOSS, but that's it, the computer is still running non-free software...
@Orca @Codeberg with "AI" being a hype/marketing/bs term, I guess everyone gets their own interpretation of it but the underlying tech, neural networks, have enabled/improved a lot of OSS applications like tesseract OCR becoming really good at text recognition from using them, years before the AI hype (still running locally with about the same resource usage). Another example could be whisper, allowing for high quality local transcription "AI".
Plenty of OSS gen AI projects too but who cares^^