Andreas Jansson

@threepointone ah yes of course it is lol

@allnoteson my guy’s implementing the actor model!

tl;dr - subagent behaviour working on adding multi chat and subagents to the agents starter (yay!) and I have a curious product direction/question. our subagents can be full fledged chats themselves. which means they could not only be async while they work on their thing and you continue, but you could continue "talking" with them after they've "returned" a result. so what should the default behaviour in the starter be? - readonly, no input. this is what most (all?) products/devtools like this do atm - have chat, but it's only followups, doesn't affect the main chat - add a "send back/summarize to main chat" this feels powerful and underexplored I'll probably ship option 1 for now, but there's something here... anyway, multichat/subagents in starter template coming this week

There’s an analog infinity between every pair of pixels

@alex__codes @abhinav17800 We immediately hit another snag after having fixed the first one. It really should work now, please shout if you still run into issues.

@abhinav17800 @allnoteson Yep same here, eventually it did fail, but this is also using an H100 GPU

My all time favorite text-to-video model is still Hunyuan Video at 320x240px.

loras are so back. wrote this blog 2 years ago.. more lora support soon blog.cloudflare.com/fine-tuned-inf…

@abhinav17800 @alex__codes I believe this should be resolved now. I'm sorry again for this!

@allnoteson @alex__codes Thanks for update , we tried to use the trainer but it’s in cold status showing as queued .

I'll be speaking at Cloudflare Immerse in Stockholm next week. Still some spots left if you're in town, should be fun. fieldmarketing.www.cloudflare.com/immersestockho…

I don't remember ever doing a talk in Sweden before? Maybe something internal at Spotify. I just hope the well dressed Stockholm crowd doesn't judge my hillbilly Småland accent.

@alex__codes @abhinav17800 Really sorry about the wait, we're still actively investigating this.

@abhinav17800 @allnoteson Any updates on this?

@abhinav17800 Thanks for reaching out, looking into this

@allnoteson Hi @allnoteson we’ve been unable to complete training on Fast Flux Trainer due to repeated errors for the last 2 days. We already reached out to support but haven’t received a response yet. Could you please help look into this issue? replicate.com/replicate/fast…

I see a fellow harnessmaxxer @allnoteson

You can now create a new @CloudflareDev account and sign up for R2 and create a bucket, buy a domain and so much more ALL from the command line courtesy of @stripe Projects Have your agent try it today 😉 blog.cloudflare.com/agents-stripe-…

Maybe it's my religious upbringing, all my favorite sci-fi books finish up in Teilhard de Chardin's noosphere. Star Maker, Childhood's End, Accelerando, Death's End... Just finished Blood Music which was fantastic. What else is there?

⚡ Your thoughts, vectorized. Move from raw concept to scalable SVG instantly with our AI SVG Generator. Infinite resolution, zero friction, total control. Available in Studio AI now: studio.creativefabrica.com/ai-svg-generat…

@philkyprianou @replicate That's strange, the safety checker is trigger happy but it's been working for me with BytePlus URLs. Could you DM me a prediction ID?

@replicate Impossible to use AI human generated image: Prediction failed: Async prediction failed: ModelError: The input or output was flagged as sensitive. Please try again with different inputs. (E005) (uIJ6l3ruRD)

Seedance 2.0 now supports consistent characters. 1. Generate character images with Seedream 5 (Lite) and set return_byteplus_urls=true 2. Use those image URLs as reference images or init images in Seedance 2.0 / Seedance 2.0 (Fast)

@CodyBaker_xx @replicate Yes I also hope that will be supported at some point. Reference videos should work though, unless they contain a human. Is it not working for you?

@replicate People want to upload their own photos. Plz @replicate & input videos. Current state of Seedance 2.0 on replicate is broken

@0xMomo @KyoeiFWD @replicate That was my first thought too haha, but no it only works with text-to-image generations

@allnoteson @KyoeiFWD @replicate Can you just upload your face as reference to seedream and tell it to “clean” it up and use that as the reference image for seedance?