Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

This smalldiffusion project is really fun. It is a tiny library for training and sampling from diffusion models. I love efforts like this, they are much easier to play around with than big libraries like diffusers. Link:

alec_helbling's profile picture

Alec Helbling

10,779 subscribers

18,844 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

I'm really having fun with rigs, I just love how it makes the whole process far easier Also this is a WIP for an animation I'd like to make :)
0:14

I'm really having fun with rigs, I just love how it makes the whole process far easier Also this is a WIP for an animation I'd like to make :)

PurpleTourmaple - (Comms Open) 🔞

10,380 Aufrufe • vor 2 Jahren

Massive update for AI Engineers! Training diffusion models just got a lot easier. dLLM is an open-source library that does for diffusion models what Hugging Face did for transformers. Here's why this matters: Traditional autoregressive models generate text left-to-right, one token at a time. Diffusion models work differently - they refine the entire sequence iteratively, giving you better control over generation quality and more flexible editing capabilities. The problem? Building and training these models required stitching together scattered tools and reimplementing research papers from scratch. Most of the tooling has been scattered and is hard to reproduce. dLLM changes this: It unifies everything you need to train, evaluate, and deploy diffusion language models: ↳ Scalable training with LoRA, DeepSpeed, and FSDP support ↳ Unified evaluation that abstracts away inference complexity ↳ Ready-to-use recipes for pretraining, finetuning, and evaluation The library includes implementations of models like LLaDA and Dream, plus training algorithms like Edit Flows that enable insertion, deletion, and substitution operations. The team just released ModernBERT-Chat models showing you can turn BERT into lightweight chatbots through masked instruction tuning. This is practical and worth exploring. The setup is straightforward. Run locally with Accelerate or scale to multi-node clusters with Slurm. If you're working with language models and want to explore diffusion-based approaches without rebuilding infrastructure, dLLM gives you a production-ready starting point. Link to the repo in the next tweet.
0:14

Massive update for AI Engineers! Training diffusion models just got a lot easier. dLLM is an open-source library that does for diffusion models what Hugging Face did for transformers. Here's why this matters: Traditional autoregressive models generate text left-to-right, one token at a time. Diffusion models work differently - they refine the entire sequence iteratively, giving you better control over generation quality and more flexible editing capabilities. The problem? Building and training these models required stitching together scattered tools and reimplementing research papers from scratch. Most of the tooling has been scattered and is hard to reproduce. dLLM changes this: It unifies everything you need to train, evaluate, and deploy diffusion language models: ↳ Scalable training with LoRA, DeepSpeed, and FSDP support ↳ Unified evaluation that abstracts away inference complexity ↳ Ready-to-use recipes for pretraining, finetuning, and evaluation The library includes implementations of models like LLaDA and Dream, plus training algorithms like Edit Flows that enable insertion, deletion, and substitution operations. The team just released ModernBERT-Chat models showing you can turn BERT into lightweight chatbots through masked instruction tuning. This is practical and worth exploring. The setup is straightforward. Run locally with Accelerate or scale to multi-node clusters with Slurm. If you're working with language models and want to explore diffusion-based approaches without rebuilding infrastructure, dLLM gives you a production-ready starting point. Link to the repo in the next tweet.

Akshay 🚀

64,011 Aufrufe • vor 11 Monaten

Assuming the boards the winds/waves protags are holding are for crossing the water, I REALLY hope you get pulled around by your pokemon like this, when I was a kid I thought it looked like so much fun 😭
0:18

Assuming the boards the winds/waves protags are holding are for crossing the water, I REALLY hope you get pulled around by your pokemon like this, when I was a kid I thought it looked like so much fun 😭

tee ꒰ა ♡ ໒꒱

230,576 Aufrufe • vor 8 Monaten

Aaaah this new wireless setup is just so much easier to play around and be a doofus in XD
0:15

Aaaah this new wireless setup is just so much easier to play around and be a doofus in XD

illumi-NATION Cyberwear

118,671 Aufrufe • vor 2 Jahren

This is surprisingly quite hard to do via APIs, but local models (like H3) have really good LoRAs, which are also trainable locally. It seems much more creativity is possible with local video models, even if they're less intelligent than API-based ones
0:10

This is surprisingly quite hard to do via APIs, but local models (like H3) have really good LoRAs, which are also trainable locally. It seems much more creativity is possible with local video models, even if they're less intelligent than API-based ones

anabology

40,550 Aufrufe • vor 16 Tagen

im already so very much in love with this part and this outfit is not making it easier for me
0:15

im already so very much in love with this part and this outfit is not making it easier for me

alice

10,984 Aufrufe • vor 10 Monaten

this is what i love so much about the spider-verse movies because like, they didn’t need to add that, it’s so TINY but they did anyways
0:14

this is what i love so much about the spider-verse movies because like, they didn’t need to add that, it’s so TINY but they did anyways

trix ♡

765,604 Aufrufe • vor 5 Monaten

AudioLDM is a text-to-audio model by Haohe Liu et al. It is available in the 🧨 Diffusers library from v0.15.0 onwards, and is capable of generating realistic music, sound effects and speech Try it in this Colab to generate music for yourself:
0:10

AudioLDM is a text-to-audio model by Haohe Liu et al. It is available in the 🧨 Diffusers library from v0.15.0 onwards, and is capable of generating realistic music, sound effects and speech Try it in this Colab to generate music for yourself:

Sanchit Gandhi

27,143 Aufrufe • vor 3 Jahren

Diffusion models are great, but we can squeeze out so much more from them. The only problem is that it usually requires extra training or manual representation editing. In our new paper, we show that with the current capabilities of LLMs, it is much simpler than we thought!
0:17

Diffusion models are great, but we can squeeze out so much more from them. The only problem is that it usually requires extra training or manual representation editing. In our new paper, we show that with the current capabilities of LLMs, it is much simpler than we thought!

Yossi

36,236 Aufrufe • vor 7 Monaten

“One in a Million“ These can hurt if they are too big. I think I like them a little softer. Big isn't always better. Big ones are easier to catch. Size matters when they are coming FAST. Not so FAST!!! One of these is really big and hard 😱
0:12

Sensitive content

“One in a Million“ These can hurt if they are too big. I think I like them a little softer. Big isn't always better. Big ones are easier to catch. Size matters when they are coming FAST. Not so FAST!!! One of these is really big and hard 😱

Little Miss Mari

14,734 Aufrufe • vor 14 Tagen

✦ MoonBlessing Aura - Free Release ✦ My biggest VFX project (so far) is finished! I would like to thank everyone for their patience, and I really hope you will like it! It has a lot of options that I tried to show as much as possible in the previews! Link in replies!
0:14

✦ MoonBlessing Aura - Free Release ✦ My biggest VFX project (so far) is finished! I would like to thank everyone for their patience, and I really hope you will like it! It has a lot of options that I tried to show as much as possible in the previews! Link in replies!

Zedkiel

109,238 Aufrufe • vor 1 Jahr

These can hurt if they are too big I think I like them a little softer Big isn't always better Big ones are easier to catch Size matters when they are coming FAST Not so FAST!!! One of these is really big and hard 😱 #usa #karategirl #martialarts #japanesegirl #asiangirl
0:15

These can hurt if they are too big I think I like them a little softer Big isn't always better Big ones are easier to catch Size matters when they are coming FAST Not so FAST!!! One of these is really big and hard 😱 #usa #karategirl #martialarts #japanesegirl #asiangirl

Little Miss Mari

11,566 Aufrufe • vor 1 Monat

Andrej Karpathy calls AI Agents slop "Overall, the models they are not there. And I feel like the industry [...] it's making too big of a jump and it's trying to pretend that this is amazing. And it's not—it's slop! And I think they are not coming to terms with it. And maybe they are trying to fundraise or something like that, I'm not sure what's going on."
0:16

Andrej Karpathy calls AI Agents slop "Overall, the models they are not there. And I feel like the industry [...] it's making too big of a jump and it's trying to pretend that this is amazing. And it's not—it's slop! And I think they are not coming to terms with it. And maybe they are trying to fundraise or something like that, I'm not sure what's going on."

Lisan al Gaib

3,275,150 Aufrufe • vor 1 Jahr

this is what a real gamer looks like (i only play league and sims 4…project zomboid sometimes…)
0:15

this is what a real gamer looks like (i only play league and sims 4…project zomboid sometimes…)

≽(•⩊ •マ≼ em💢

13,090 Aufrufe • vor 1 Monat

my favourite little detail about the game segments is that your raft movements have a subtle trail effect to look like burn-in all the tiny bits done for these board parts like this really added so much to the experience and I really appreciate the work that was put into it
0:12

my favourite little detail about the game segments is that your raft movements have a subtle trail effect to look like burn-in all the tiny bits done for these board parts like this really added so much to the experience and I really appreciate the work that was put into it

🍉 guru 🍉

56,338 Aufrufe • vor 1 Jahr

gosh i love this tiny domestic moment so much like you can see how ilya struggles with not shutting down from shane how he speaks to him way softer than ever before how he reassures him after shane apologizes, how he caresses his knee my god they are so in love it hurts
0:15

gosh i love this tiny domestic moment so much like you can see how ilya struggles with not shutting down from shane how he speaks to him way softer than ever before how he reassures him after shane apologizes, how he caresses his knee my god they are so in love it hurts

ellie

96,437 Aufrufe • vor 6 Monaten

Windy, the feral mare, has avoided predators like this cougar wandering through her area for 28yrs. This cat is only 50yds from her waterer but for a cougar, deer are much easier prey.
0:10

Windy, the feral mare, has avoided predators like this cougar wandering through her area for 28yrs. This cat is only 50yds from her waterer but for a cougar, deer are much easier prey.

The Thankful Outdoorsman

32,139 Aufrufe • vor 5 Monaten

I wanted to test one of the models out and so I made a chica walking cycle for the fun of it. ============= Please note that it's not suppose to be really accurate to the movie, but have that taste from the movie. THIS IS ALSO NOT RELATED TO MY PROJECT!!
0:17

I wanted to test one of the models out and so I made a chica walking cycle for the fun of it. ============= Please note that it's not suppose to be really accurate to the movie, but have that taste from the movie. THIS IS ALSO NOT RELATED TO MY PROJECT!!

UNKNOWNSPECIM3N

38,285 Aufrufe • vor 2 Jahren

i love this video so much she is so big but still so tiny 🥺
0:14

i love this video so much she is so big but still so tiny 🥺

becs

42,509 Aufrufe • vor 5 Monaten