LeapLab@THU

LeapLab@THU

[AK @_akhaliq]

discuss: https://t.co/yXpTW98QCE

[AK @_akhaliq]

Microsoft releases ART

Anonymous Region Transformer for Variable Multi-Layer Transparent Image Generation

[LeapLab@THU @LeapLabTHU]

🔬 Read the paper: https://t.co/bQ8PKWWgaG
💻 Code & pretrained models: https://t.co/ps61ivPOOZ

[LeapLab@THU @LeapLabTHU]

🚀 Excited to share our work on #ECCV2024: "AdaNAT: Exploring Adaptive Policy for Token-Based Image Generation".

🖼️ We introduce AdaNAT, a novel approach for efficient and high-quality image generation using adaptive policies in Non-autoregressive Transformers.

[LeapLab@THU @LeapLabTHU]

🔑 Key features:

Learnable policy network for adaptive modulation of token generation
Adversarial reward model for improved quality and diversity
Significantly reduced inference time compared to diffusion models
📊 Impressive results on ImageNet, MSCOCO, and CC3M datasets!

[Yang Yue @YangYue_THU]

very fantastic work done with @ALucyBrilliant @RayLu_THU @AndrewZ45732491. direct parameter editing can modulate LLM’s behavior

[AK @_akhaliq]

ConvLLaVA

Hierarchical Backbones as Visual Encoder for Large Multimodal Models

High-resolution Large Multimodal Models (LMMs) encounter the challenges of excessive visual tokens and quadratic visual complexity. Current high-resolution LMMs address the quadratic

[Chunjiang Ge @GeChunjiang]

📢Excited to share our recent work on Large Multimodal Models: ConvLLaVA. Without the encoding multiple image patches and multiple encoders, we use a hierarchical backbone, ConvNeXt, realizing high resolution understanding.

https://t.co/eIEYDE76mV

[LeapLab@THU @LeapLabTHU]

EfficientTrain++ is accepted by TPAMI2024🤩

🔥An off-the-shelf, easy-to-implement algorithm for training foundation visual backbones efficiently!
🔥1.5−3.0× lossless training/pre-training speedup on ImageNet-1K/22K!
Paper&Code:
https://t.co/Fcs9APIQQK
https://t.co/jsddWSIqAw

[LeapLab@THU @LeapLabTHU]

EfficientTrain++ is accepted by TPAMI2024🤩

🔥An off-the-shelf, easy-to-implement algorithm for training foundation visual backbones efficiently!
🔥1.5−3.0× lossless training/pre-training speedup on ImageNet-1K/22K!
Paper&Code:
https://t.co/Fcs9APIQQK
https://t.co/jsddWSIqAw

[LeapLab@THU @LeapLabTHU]

Our recent work: Agent Attention!
[High Performance & Linear Complexity]
[Double the speed of SD and enhance generation quality, no additional fine-tuning is required]
Paper and code:
https://t.co/kejcBAtWJC
https://t.co/REDUipdUEw
https://t.co/wFAXNsY3qo

[LeapLab@THU @LeapLabTHU]

Excited to share our #NeurIPS2023 spotlight paper! 🌟 It proposes a novel offline-to-online RL algorithm, efficiently utilizing collected samples by training a family of policies offline and selecting suitable ones online. Check out our paper for details! https://t.co/bEq76wa6MA

[Andrew Zhao @_AndrewZhao]

ExpeL is now accepted at #AAAI24 ! The code and camera ready version will be updated promptly. Thanks for all the collaborators and see you in Vancouver!

[Rui Lu @RayLu_THU]

Check us out at #NeurIPS2023 poster！We investigate into Q-value divergence phenomenon in offline RL and find self-excitation to be the main reason. Using layernorm in RL models can fundamentally prevent this from happening. https://t.co/Pqs7JbG77n

[LeapLab@THU @LeapLabTHU]

Our recent work: Agent Attention!

[High Performance & Linear Complexity]
[Accelerate and improve Stable Diffusion, no additional fine-tuning is required]

The paper and code have been released:
https://t.co/REDUipdUEw
https://t.co/wFAXNsY3qo