#4o — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #4o, aggregated by home.social.
-
Milers a Jardinets #FreePalestine #4O Aturem el Genocidi!
-
Columna andreuenca per la mani 4-O
La Cinètika, dissabte, 4 d’octubre, a les 10:30 CEST
El proper dissabte 4 d’octubre, Sant Andreu del Palomar se sumarà a la jornada de mobilització convocada per exigir la fi del comerç d’armes i la ruptura de les relacions amb l’Estat d’Israel.
Per aquest motiu, des de La Cinètika fem una crida a totes les persones, entitats i col·lectius del poble a participar en una columna unitària que es dirigirà cap a la manifestació central, convocada el mateix dia a les 12h als Jardinets de Gràcia, a Barcelona.
Ara més que mai, cal articular una resposta clara i decidida davant el genocidi perpetrat per les forces imperialistes a Palestina i expressar, amb fermesa, el nostre rebuig al terrorisme d’Estat exercit per Israel, al projecte sionista i als seus aliats internacionals.
La Columna de Sant Andreu – 4 d’octubre. Punt de trobada a Fabra i Puig, 28 (La Cinètika) a les 10:30h. A les 11h serem a la xurreria del Clot per anar plegades amb les companyes de la Zona Nord de Barcelona en columna cap a Barcelona.
https://bcn.convoca.la/event/columna-andreuenca-per-la-mani-4-o
-
Manifestació del 4 d’octubre ✊
Fi al comerç d’armes i a les relacions amb Israel📍 Barcelona, Jardinets de Gràcia
📅 4 d’octubre ⏰ 12 hManifestació unitària amb la participació de la Comunitat Palestina de Catalunya i Prou Complicitat amb Israel, en el marc de la crida estatal de la Red Solidaria Contra la Ocupación de Palestina (RESCOP), més informació: https://bsky.app/.../rescop.bsky.social/post/3lzavveyqqs2a.
#FreePalestine #Palestina #StopGenocide #StopTheGenocide #Gaza #SaveGaza #MareaPalestina #PalestinaLliure #4O
-
-
The latest version of #ChatGpt #4o is very shit at coding.
I can see why #vibecode folk are frustrated and #luddites rejoice.Eg;
Context "forgets" key spec points of the session, even ones that are critical to ops.You have to jam everything with special checkpoint directives and hammer specs in, least they get forgotten.
It's terrible (and I have used earlier versions and #Claude)
In my opinion, what is happening because of the high compute burn, #OpenAi optimised the session manager. It burns less compute but is also more shit. The balance must be restored.
-
More you ask ChatGPT to render images in the same conversation with each new render it will make image look worst, but I found that if you upload previously rendered first image in a new chat and ask to keep the style it works way better.
-
-
Not much is known about the internal working of #ChatGPT's new #4o image generation model. What seems clear is that the latent representation is dependent on orientation. If asked to render things upside-down, it struggles. #Synthography
-
CW: Somewhat frightening animal character
#ChatGPT #4o renders a realistic version of a Minecraft creeper. #Synthography
-
298: BigQuery Gits it With Devops Say “hello” in whatever funny accent you can think of to the future of voice agents! OpenAI’s new GPT-4o Transcribe models are setting a new benchmark in speech-to-text accuracy. Noisy WeWork office? Cubicle neighbor listening to The Cloud Pod? It’s no match for 4o! #chatgpt #4o #thecloudpod https://www.thecloudpod.net/?p=21037
-
298: BigQuery Gits it With Devops Say “hello” in whatever funny accent you can think of to the future of voice agents! OpenAI’s new GPT-4o Transcribe models are setting a new benchmark in speech-to-text accuracy. Noisy WeWork office? Cubicle neighbor listening to The Cloud Pod? It’s no match for 4o! #chatgpt #4o #thecloudpod https://www.thecloudpod.net/?p=21037
-
I was having fun with ChatGPT as a replacement tool for Photoshop / Blender. My goal wasn't to make a perfect replacement for these icons, I was doodling with the system to see how it performs different tasks. The final version still needs some Photoshop to polish the pixels. These icons are the best looking ones, some were really bad or I didn't achieve my goal at all.
It's cool that it faked LeBron James signature and others.
-
Another attempt to modernize old website by using ChatGPT.
-
Updating old Apple website to a modern devices with ChatGPT. Cute! The layout is still legit, still can be used even today.
-
-
Native image generation in #ChatGPT #4o opens up a range of new possibilities, for example, using style references. Here are the results of the same request, “box art for the game The Long Dark, reimagined as a black box NES game”, the first with just the text prompt, and the second with an example box art image. As you can see, ChatGPT kind of has a general idea of what a black box game is, but is lacking in detail. In the second, it basically nails it, with just some minor flaws in the text.
-
OpenAI veröffentlicht Best Practices für Reasoning-Modelle
OpenAI hat eine ausführliche Anleitung für die optimale Nutzung ihrer Reasoning-Modelle (z. B. o1 und o3-mini) im Vergleich zu den klassischen GPT-Modellen veröffentlicht. Der Leitfaden liefert Informationen darüber, wann und wie die Modelle am besten eingesetzt werden können, um
https://www.apfeltalk.de/magazin/news/openai-veroeffentlicht-best-practices-fuer-reasoning-modelle/
#News #Services #4Mini #4o #GPT #OpenAI #Reasoning -
Is #ChatGPT censoring contents on the US presidency? And #DeepSeek? How do these AIs value #democracy? Well, you can get ChatGPT #4o to say that #Trump is a bad president for democracy, but with Xi Jinping and DeepSeek? Not that easy... #censorship #Openai #Xijinping #china #USA #AI #LLM #Capitolhill #pardon #hongkong #umbrella #democracy #AItest
-
In need for an equivalent of #OpenAI 's o1, latest OpenAI's model better than #GPT - #4o , but you would like it to be @OpenSource ?
#DeepSeek is what you need , and it is #chinese .
( #DeepSeek-R1-Lite-Preview is out)
It is directly competing with #GPT - #o1 , last year's model that overcame 83% of complex mathematical problems , and that overcame #GPT - #4o .#DeepSeek also openly exposés its #ChainOfThink ( #CoT )
-
Today I spent quite some time with #chatgpt #ai on its model #4o , to try and generate some #SVG markup for some images I needed for a web application.
#Graphicdesign ers: you’re safe. The likes of ChatGPT and similar #LLM - based a.i.-s simply can’t compete with a graphic artist using #inkscape to create some SVG.
What I wanted: a spiral arrow pointing right, tapered from narrow to thick.
What I got: bull shyte.
