EVA-02: A Visual Representation for Neon Genesis
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:MR5VZR5Rrecord.jsonopen to challenge →
read the original abstract
We launch EVA-02, a next-generation Transformer-based visual representation pre-trained to reconstruct strong and robust language-aligned vision features via masked image modeling. With an updated plain Transformer architecture as well as extensive pre-training from an open & accessible giant CLIP vision encoder, EVA-02 demonstrates superior performance compared to prior state-of-the-art approaches across various representative vision tasks, while utilizing significantly fewer parameters and compute budgets. Notably, using exclusively publicly accessible training data, EVA-02 with only 304M parameters achieves a phenomenal 90.0 fine-tuning top-1 accuracy on ImageNet-1K val set. Additionally, our EVA-02-CLIP can reach up to 80.4 zero-shot top-1 on ImageNet-1K, outperforming the previous largest & best open-sourced CLIP with only ~1/6 parameters and ~1/6 image-text training data. We offer four EVA-02 variants in various model sizes, ranging from 6M to 304M parameters, all with impressive performance. To facilitate open access and open research, we release the complete suite of EVA-02 to the community at https://github.com/baaivision/EVA/tree/master/EVA-02.
This paper has not been read by Pith yet.
Forward citations
Cited by 17 Pith papers
-
Spatially Prompted Visual Trajectory Prediction for Egocentric Manipulation
The paper introduces SP-VTP as a new setting for egocentric manipulation, releases the EgoSPT dataset with first-frame spatial annotations, and proposes the SPOT model that outperforms non-prompted baselines on cross-...
-
Unlocking Compositional Generalization in Continual Few-Shot Learning
A dual-phase framework using self-supervised ViT slots optimizes representations for class identity during training and composes them dynamically at inference to achieve state-of-the-art generalization to unseen conce...
-
Closed-Form Linear-Probe Dataset Distillation for Pre-trained Vision Models
CLP-DD distills small synthetic datasets for linear probing on pre-trained models via closed-form inner solver and discriminative outer loss, matching or exceeding LGM+DSA performance at much lower cost on ImageNet-10...
-
Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation
Scaled vanilla autoregressive models based on Llama achieve 2.18 FID on ImageNet 256x256 image generation, beating popular diffusion models without visual inductive biases.
-
Beyond Encoder Accumulation: Measuring Encoder Roles in Multi-Encoder VLMs
Retraining all 31 subsets of five vision encoders shows Capacity and Necessity are distinct, pre-projector effective rank predicts residual performance at fixed parameter count, and high-Capacity plus adaptive complem...
-
ClawEnvKit: Automatic Environment Generation for Claw-Like Agents
ClawEnvKit automates generation of diverse verified environments for claw-like agents from natural language, producing the Auto-ClawEval benchmark of 1,040 environments that matches human-curated quality at 13,800x lo...
-
Bias at the End of the Score
Reward models used as quality scorers in text-to-image generation encode demographic biases that cause reward-guided training to sexualize female subjects, reinforce stereotypes, and reduce diversity.
-
Hydra-MDP: End-to-end Multimodal Planning with Multi-target Hydra-Distillation
Hydra-MDP uses multi-teacher distillation and a multi-head decoder to learn diverse, metric-specific trajectories in an end-to-end autonomous-driving planner, winning the Navsim challenge.
-
EVA-CLIP: Improved Training Techniques for CLIP at Scale
EVA-CLIP delivers improved CLIP training recipes that yield 82.0% zero-shot ImageNet-1K accuracy for a 5B-parameter model after only 9 billion samples.
-
Unlocking Compositional Generalization in Continual Few-Shot Learning
A decoupling strategy optimizes object slots for holistic class identity during training and composes them at inference to achieve better generalization to unseen concepts in continual few-shot settings.
-
Weak-to-Strong Knowledge Distillation Accelerates Visual Learning
Weak-to-strong knowledge distillation applied early and then turned off accelerates convergence to target performance in visual learning tasks by factors of 1.7-4.8x.
-
SMFormer: Empowering Self-supervised Stereo Matching via Foundation Models and Data Augmentation
SMFormer achieves state-of-the-art self-supervised stereo matching by using vision foundation models for disturbance-resistant features and data augmentation to enforce output consistency, rivaling or exceeding some s...
-
InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks
InternVL scales a vision model to 6B parameters and aligns it with LLMs using web data to achieve state-of-the-art results on 32 visual-linguistic benchmarks.
-
CADS: Conformal Adaptive Decision System for Cost-Efficient Image Classification
CADS is a conformal-prediction-driven model cascade that routes images to scout or oracle models based on estimated complexity to reduce inference cost while preserving accuracy.
-
CADS: Conformal Adaptive Decision System for Cost-Efficient Image Classification
CADS is a conformal-prediction-driven adaptive system that cascades models for image classification to achieve up to 12x lower computational cost while maintaining accuracy.
-
Advancing Vision Transformer with Enhanced Spatial Priors
EVT improves Vision Transformers by using Euclidean distance decay for spatial priors and simpler grouping, achieving 86.6% top-1 accuracy on ImageNet-1k.
-
ClawEnvKit: Automatic Environment Generation for Claw-Like Agents
EVT improves the RMT backbone by using Euclidean-distance attention decay and 1D token grouping, achieving 86.6% top-1 on ImageNet-1K at 384×384 resolution.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.