HuggingFaceTB/stack-edu-prompts-16langs-1k
Viewer
•
Updated
•
1k
Loubna Ben Allal
loubnabnl
AI & ML interests
SmolLMs, ML for code, data
Recent Activity
updated
a dataset
about 15 hours ago
HuggingFaceTB/stack-edu-prompts-16langs-1k
published
a dataset
about 15 hours ago
HuggingFaceTB/stack-edu-prompts-16langs-1k
published
an
article
1 day ago
SmolVLA: Efficient Vision-Language-Action Model trained on Lerobot Community Data
Organizations
loubnabnl's activity
updated
a
dataset
about 15 hours ago
published
a
dataset
about 15 hours ago
published
an
article
1 day ago
Article
SmolVLA: Efficient Vision-Language-Action Model trained on Lerobot Community Data
By
and 8 others
•
•
47
upvoted
an
article
7 days ago
Article
CodeAgents + Structure: A Better Way to Execute Actions
By
and 1 other
•
•
37
upvoted
a
paper
8 days ago
reacted to
clem's
post with 🚀🔥
9 days ago
Post
3439
Playing with Veo3 this morning. Share your prompt if you want me to create videos for you (bonus point if they funnily reference HF/open-source). These videos are "a cat on the moon rapping "I love Hugging Face""!
- 25 replies
reacted to
nyuuzyou's
post with 🔥
9 days ago
Post
2903
I recently updated
nyuuzyou/pxhere dataset and it now contains approximately 1.1M CC0 high-resolution images
published
a
dataset
12 days ago
upvoted
3
changelogs
12 days ago
Changelog
Static Spaces can now have a build step
•
82
Changelog
Xet is now the default storage option for new users and organizations
•
53
Changelog
AI-generated Abstract summaries on Hugging Face Papers
•
65
reacted to
merve's
post with 🔥
12 days ago
Post
2574
Google released MedGemma on I/O'25 👏
google/medgemma-release-680aade845f90bec6a3f60c4
> 4B and 27B instruction fine-tuned vision LMs and a 4B pre-trained vision LM for medicine
> available with transformers from the get-go 🤗
they also released a cool demo for scan reading ➡️ google/rad_explain
use with transformers ⤵️
> 4B and 27B instruction fine-tuned vision LMs and a 4B pre-trained vision LM for medicine
> available with transformers from the get-go 🤗
they also released a cool demo for scan reading ➡️ google/rad_explain
use with transformers ⤵️
- 1 reply
replied to
their
post
14 days ago
it does now :)
reacted to
AdinaY's
post with 🔥🚀
14 days ago
Post
2765
ByteDance is absolutely cooking lately🔥
BAGEL 🥯 7B active parameter open multimodal foundation model by Bytedance Seed team.
ByteDance-Seed/BAGEL-7B-MoT
✨ Apache 2.0
✨ Outperforms top VLMs (Qwen2.5-VL & InternVL-2.5)
✨ Mixture-of-Transformer-Experts + dual encoders
✨ Trained on trillions of interleaved tokens
BAGEL 🥯 7B active parameter open multimodal foundation model by Bytedance Seed team.
ByteDance-Seed/BAGEL-7B-MoT
✨ Apache 2.0
✨ Outperforms top VLMs (Qwen2.5-VL & InternVL-2.5)
✨ Mixture-of-Transformer-Experts + dual encoders
✨ Trained on trillions of interleaved tokens
reacted to
sayakpaul's
post with 🔥
14 days ago
Post
1672
Despite the emergence of combining LLM and DiT architectures for T2I synthesis, its design remains severely understudied.
This was done long ago and got into CVPR25 -- super excited to finally share it now, along with the data and code ♥️
We explore several architectural choices that affect this design. We provide an open & reproducible training recipe that works at scale.
Works like Playground v3 have already explored a deep fusion between an LLM and a DiT, sharing their representations through layerwise attention. They exhibit excellent performance on T2I.
Despite its compelling results and other performance virtues, it remains unexplored, which is what we want to improve in our work. Specifically, we take a pre-trained LLM (Gemma-2B) and trainable DiT, and set out to explore what makes a "good deep fusion" between the two for T2I.
We explore several key questions in the work, such as:
Q1: How should we do attention? We considered several alternatives. PixArt-Alpha like attention (cross-attention) is very promising.
Q2: Should we incorporate additional text modulation?
Q3: Can we eliminate timestep conditioning?
Q4: How do we do positional encodings?
Q5: Do instruction-tuned LLMs help deep fusion?
Q6: Would using a decoder LLM from a multimodal model be helpful?
Q7: Does using a better variant of Gemma help?
Based on the above findings, we arrive at FuseDiT with the following components on top of the base architecture from the findings of our experiments.
* No AdaLN-Zero modules
* 1D + 2D-RoPE
* Gemma 2 2B, adjusting DiT configurations accordingly
We trained FuseDiT on a mixture from CC12M, JourneyDB, & SA (~26M image-text pairs) for 800 steps. While not the best model, it's encouraging to develop something in a guided manner using open datasets.
To know more (code, models, all are available), please check out the paper:
https://lnkd.in/gg6qyqZX.
This was done long ago and got into CVPR25 -- super excited to finally share it now, along with the data and code ♥️
We explore several architectural choices that affect this design. We provide an open & reproducible training recipe that works at scale.
Works like Playground v3 have already explored a deep fusion between an LLM and a DiT, sharing their representations through layerwise attention. They exhibit excellent performance on T2I.
Despite its compelling results and other performance virtues, it remains unexplored, which is what we want to improve in our work. Specifically, we take a pre-trained LLM (Gemma-2B) and trainable DiT, and set out to explore what makes a "good deep fusion" between the two for T2I.
We explore several key questions in the work, such as:
Q1: How should we do attention? We considered several alternatives. PixArt-Alpha like attention (cross-attention) is very promising.
Q2: Should we incorporate additional text modulation?
Q3: Can we eliminate timestep conditioning?
Q4: How do we do positional encodings?
Q5: Do instruction-tuned LLMs help deep fusion?
Q6: Would using a decoder LLM from a multimodal model be helpful?
Q7: Does using a better variant of Gemma help?
Based on the above findings, we arrive at FuseDiT with the following components on top of the base architecture from the findings of our experiments.
* No AdaLN-Zero modules
* 1D + 2D-RoPE
* Gemma 2 2B, adjusting DiT configurations accordingly
We trained FuseDiT on a mixture from CC12M, JourneyDB, & SA (~26M image-text pairs) for 800 steps. While not the best model, it's encouraging to develop something in a guided manner using open datasets.
To know more (code, models, all are available), please check out the paper:
https://lnkd.in/gg6qyqZX.
liked
a
Space
14 days ago
liked
2
datasets
16 days ago