Releasing weights for FLUX.1 Krea • Hacker News Board

Re: Releasing weights for FLUX.1 Krea

#51

Post by dvrp » Thu, Jul 31, 2025, 10:33 PM UTC

jackphilson wrote: post #47

dvrp wrote: post #2

Hello everyone. I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time. My team and I will try to be online and try to answer any questions you may have throughout the day.

Hi. Thanks for this. What is your goal of doing so? From a business standpoint. Or is it purely altruistic?

Haha-classic!

It’s simple: hackability and recruiting!

The open-source community hacking around it and playing with it PLUS talented engineers who may be interested in working with us already makes this release worth it. A single talented distributed systems engineer has a lot of impact here.

Also, the company ethos is around AI hackability/controllability, high-bar for talent, and AI for creatives - so this aligns perfectly.

The fact that Krea serves both in-house and 3rd-Party models tells you that we are not that bullish on models being a moat.

Re: Releasing weights for FLUX.1 Krea

#52

Post by nodja » Thu, Jul 31, 2025, 10:35 PM UTC

zamadatix wrote: post #28

7734128 wrote: post #21

Earlier quoted context omitted.

I've never seen a 32 bit model. There's bound to be a few of them, but it's hardly a normal precision.

Some of the most famous models were distributed as F32, e.g. GPT-2. As things have shifted more towards mass consumption of model weights it's become less and less common to see.

> As things have shifted more towards mass consumption of model weights it's become less and less common to see.

Not the real reason. The real reason is that training has moved to FP/BF16 over the years as NVIDIA made that more efficient in their hardware, the same reason you're starting to see some models being released in 8bit formats (deepseek).

Of course people can always quantize the weights to smaller sizes, but the master versions of the weights is usually 16bit.

Re: Releasing weights for FLUX.1 Krea

#53

Post by swyx » Thu, Jul 31, 2025, 10:43 PM UTC

sangwulee wrote: post #32

swyx wrote: post #25

Earlier quoted context omitted.

thanks for doing this! what does " designed to be compatible with FLUX architecture" mean and why is that important?

FLUX.1 is one of the most popular open weights text-to-image models. We distilled Krea-1 to FLUX.1 [dev] model so that the community can adopt it seamlessly into existing ecosystem. Any finetuning code, workflows, etc that was built on top of FLUX.1 [dev] can be reused with our model :)

do LoRAs conflict with your distillation?

Re: Releasing weights for FLUX.1 Krea

#54

Post by jacooper » Thu, Jul 31, 2025, 10:46 PM UTC

Non-commercial license... What's the point even, I can't use it for anything.

Re: Releasing weights for FLUX.1 Krea

#55

Post by hhh » Thu, Jul 31, 2025, 10:47 PM UTC

jacooper wrote: post #54

Non-commercial license... What's the point even, I can't use it for anything.

for the love of the game, not for corporate greed

Re: Releasing weights for FLUX.1 Krea

#56

Post by wjrb » Thu, Jul 31, 2025, 10:49 PM UTC

dvrp wrote: post #51

jackphilson wrote: post #47

Earlier quoted context omitted.

Hi. Thanks for this. What is your goal of doing so? From a business standpoint. Or is it purely altruistic?

Haha-classic! It’s simple: hackability and recruiting! The open-source community hacking around it and playing with it PLUS talented engineers who may be interested in working with us already makes this release worth it. A single talented distributed systems engineer has a lot of impact here. Also, the company ethos is around AI hackability/controllability, high-bar for talent, and AI for creatives - so this aligns p…

I can say that it's definitely working on me! I hadn't heard of Krea before, and this is a great introduction to your work. Thanks for sharing it.

Re: Releasing weights for FLUX.1 Krea

#57

Post by cchance » Thu, Jul 31, 2025, 10:59 PM UTC

wjrb wrote: post #56

dvrp wrote: post #51

Earlier quoted context omitted.

Haha-classic! It’s simple: hackability and recruiting! The open-source community hacking around it and playing with it PLUS talented engineers who may be interested in working with us already makes this release worth it. A single talented distributed systems engineer has a lot of impact here. Also, the company ethos is around AI hackability/controllability, high-bar for talent, and AI for creatives - so this aligns p…

I can say that it's definitely working on me! I hadn't heard of Krea before, and this is a great introduction to your work. Thanks for sharing it.

People underestimate how much goodwill companies gain from pushing opensource stuff out, not just from word of mouth but even picking up users for their commercial offerings too, while i could run opensource and appreciate it in a lot of cases using API's from the companies that i like (mostly ones that do opensource stuff) tends to be easier for bigger stuff...

Re: Releasing weights for FLUX.1 Krea

#58

Post by mk_stjames » Thu, Jul 31, 2025, 11:05 PM UTC

dvrp wrote: post #2

Hello everyone. I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time. My team and I will try to be online and try to answer any questions you may have throughout the day.

Any plans to get into working with the Flux 'Kontext' version, the editing models? I think the use cases of such prompted image editing is just wildly huge. Their demo blew my mind, although I haven't seen the quality of the open weight version yet. It is also a 12B distill.

Re: Releasing weights for FLUX.1 Krea

#59

Post by yieldcrv » Thu, Jul 31, 2025, 11:16 PM UTC

cchance wrote: post #57

wjrb wrote: post #56

Earlier quoted context omitted.

I can say that it's definitely working on me! I hadn't heard of Krea before, and this is a great introduction to your work. Thanks for sharing it.

People underestimate how much goodwill companies gain from pushing opensource stuff out, not just from word of mouth but even picking up users for their commercial offerings too, while i could run opensource and appreciate it in a lot of cases using API's from the companies that i like (mostly ones that do opensource stuff) tends to be easier for bigger stuff...

(unless the code repository and history is embarrassingly bad, which is most repositories)

Re: Releasing weights for FLUX.1 Krea

#60

Post by sangwulee » Thu, Jul 31, 2025, 11:20 PM UTC

swyx wrote: post #53

sangwulee wrote: post #32

Earlier quoted context omitted.

FLUX.1 is one of the most popular open weights text-to-image models. We distilled Krea-1 to FLUX.1 [dev] model so that the community can adopt it seamlessly into existing ecosystem. Any finetuning code, workflows, etc that was built on top of FLUX.1 [dev] can be reused with our model :)

do LoRAs conflict with your distillation?

The architecture is the same so we found that some LoRAs work out-of-the box, but some LoRAs don't. In those cases, I would expect people to re-run their LoRA finetuning with the trainer they've used.