←
apolinario (poli)4,771 posts
apolinario (poli) avatar✓

apolinario (poli) ✓

@multimodalart

ML Engineer for Art and Creativity @HuggingFace

◷ Joined July 2021▧ 1,168 Photos and videos
Victor M avatar
⟳ apolinario (poli) retweeted
Victor M@victormustar·Nov 17

LongCat Avatar 1.6 dropped on Hugging Face and it looks great for talking-avatar (probably SOTA) + MIT license 🔥 Made a Hugging Face Space demo for it: > input: image + audio > output: talking avatar video Lots of cool things to imagine and make with it imo. You can just point your agents to the HF url and start building with it btw 👇

0:21
Hugging Apps avatar
⟳ apolinario (poli) retweeted
Hugging Apps@HuggingApps·Nov 16

Someone trained a GTA world model from scratch before GTA VI 🤯 starting from one screenshot, hold WASD, frames are all generated live, ~17 fps on a single RTX 4060 Ti, 30 fps on better gpus ▶️ streamed to your browser on @huggingface hf.co/spaces/lucidml/gta

0:28
Linoy Tsaban avatar
⟳ apolinario (poli) retweeted
Linoy Tsaban@linoy_tsaban·Nov 16

The things we can do when models are open sourced🔥 Meet Open Vision Banana 🍌 > Flux 2 klein 4B/9B finetunes, trained to make it ✨ see ✨ > fully open recipe for generative visual perception > 1 model, 7 vision and generation tasks Space: huggingface.co/spaces/huggin… Model: huggingface.co/FunAILab/Open… Code: github.com/OpenVisionBanana/…

0:19
Aravind Srinivas avatar
⟳ apolinario (poli) retweeted
Aravind Srinivas@AravSrinivas·Nov 12

We're open-sourcing our latest versions of Decision Models: Decider v1.3 and Decider Flash v1.3, and making them free on @OpenRouter for a week. The new capability in v1.3 models is "locate". This lets you describe what to find, and the model returns the matching text with character offsets. This is useful for batch-mode decision + evidence/entity extraction. On 17 out-of-domain NER, PII and extractive QA tasks, Decider models significantly outperform the best general extraction model. And, on 63 of our own production classification tasks, v1.3 is 6.54 points more accurate than the classifiers we run today. Weights are open on Hugging Face: perplexity.ai/hub/blog/perpl…

0:16
Linoy Tsaban avatar
⟳ apolinario (poli) retweeted
Linoy Tsaban@linoy_tsaban·Nov 11

a little late to the party BUT I think H3-Image might be my new favorite text-to-image model 🔥 > capable of 16 MP generation > fast & flexible > great realism, prompt following > bonus: works with H3 2-step LoRA for realtime inference on ZeroGPU⚡️⚡️ model: huggingface.co/MiniMaxAI/Min… space: huggingface.co/spaces/MiniMa… realtime space: huggingface.co/spaces/akhali…

0:20
Hugging Apps avatar
⟳ apolinario (poli) retweeted
Hugging Apps@HuggingApps·Nov 10

Official @MiniMax_AI H3 Image just landed on Hugging Face 🎨 Generate or edit images (up to reference 8 images) in just 6 steps. 4K generation supported been a while since an image model didn't excite me this much! the aesthetics are fresh! ▶️ hf.co/spaces/MiniMaxAI/minim…

0:43
MiniMax (official) avatar
⟳ apolinario (poli) retweeted
MiniMax (official)@MiniMax_AI·Nov 10

4K image generation in 10 seconds! ⚡ Meet MiniMax H3-Image. 16 million pixel resolution. Stunning realism. Precise prompt following & editing. Up to 16 reference images. All in 10 seconds. And it's open-weight. 🚀 Available now via our API, MiniMax Code, and Hailuo AI — with 50% off 16MP+ generations during launch week! Can't wait to see what you create. 🔥 Hugging Face: huggingface.co/MiniMaxAI/Min… API: platform.minimax.io/docs/gui… MiniMax Code: code.minimax.io/about

0:50
Lysandre avatar
⟳ apolinario (poli) retweeted
Lysandre@LysandreJik·Nov 10

A pretty important new primitive just landed in 🤗 Transformers: **decision models**. I think this could matter much more than adding support for any single model. Today, different decision models all approach "making a decision" differently: - @typesafeai’s Jev, the OG: typed decisions + calibrated probabilities, single forward - @AutoTrustAI’s JEV builds open System One models around a learned decision block, w/ multimodal. - @jaredpalmer’s KEV uses a pointer head over the options, sharing the computation of the state across questions. - @Cloudflare’s CLEF uses a joint schema head that routes evidence between the state, questions and options, scoring the whole schema together. - @perplexity_ai’s Decider uses a separate 255-way readout on top of Qwen, with non-causal attention in its full-attention layers. - @liquidai’s d1 is exploring small, multimodal decision models that can run all the way down to the edge. - And @XunzhuoLiu's Semantic Router team is pushing the serving side hard with System One + the D3 family: one typed interface for choice, score and yes/no decisions, with models from sub-1B to 27B. That experimentation is good, and needed! But without a common interface, every training framework, inference engine and evaluation tool has to understand every implementation separately. So Transformers is introducing a shared decision-model layer: `DecisionConfig` `pipeline("decision")` and, most importantly: `model.decide(...)` The idea is **not** to make every decision model work the same internally, but for them to speak the same language to the ecosystem around them. This means that, without running custom or remote code: - inference engines can serve them through one interface - training tools can build against common primitives. - the Hub can package them consistently. - evaluations can compare them without model-specific glue everywhere. This is the part of transformers I'm the most excited about: to grow alongside the ecosystem and the many ideas; and to make all of them more usable by everyone. github.com/huggingface/trans…

Post attachment
apolinario (poli) avatar
apolinario (poli)@multimodalart·Nov 3

Pruna fine-tuned H3 to spit out a single frame, converting a video model back to an image model and it does have a breath of fresh air in the aesthetics, I'm getting outputs out of this model that I can't get from other models

Post attachmentPost attachment
0:29
Linoy Tsaban avatar
⟳ apolinario (poli) retweeted
Linoy Tsaban@linoy_tsaban·Nov 1

✨ Qwen Image 2.1 Realtime just got an upgrade ✨ > now using v1.5 of my 2-step distill LoRA - same speed with a quality boost > running in realtime on ZeroGPU *video not sped up* ▶️huggingface.co/spaces/linoyt…

0:16