Notice
Recent Posts
Recent Comments
Link
| 일 | 월 | 화 | 수 | 목 | 금 | 토 |
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | ||||
| 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 11 | 12 | 13 | 14 | 15 | 16 | 17 |
| 18 | 19 | 20 | 21 | 22 | 23 | 24 |
| 25 | 26 | 27 | 28 | 29 | 30 | 31 |
Tags
- prompt learning
- Mask diffusion
- Markov transition matrix
- Continual Learning
- Facial Landmark Localization
- Class Incremental Learning
- Face Pose Estimation
- learning to prompt for continual learning
- 베이지안 정리
- Img2pose
- L2P
- Energy-based model
- CVPR2022
- mmcv
- Face Alignment
- timm
- PnP algorithm
- img2pose: Face Alignment and Detection via 6DoF
- VQ-VAE
- state_dict()
- VQ-diffusion
- Vector Quantized Diffusion Model for Text-to-Image Synthesis
- requires_grad
- Discrete diffusion
- DualPrompt
- learning to prompt
- Mask-and-replace diffusion strategy
- CIL
- ENERGY-BASED MODELS FOR CONTINUAL LEARNING
- Class Incremental
Archives
- Today
- Total
목록learning to prompt (1)
Computer Vision , AI
[One-page summary] Learning to Prompt for Continual Learning (CVPR 2022) By Wang et al.
Category: continual learning (class incremental) ●Summary: Use prompt with a pre-trained model for rehearsal buffer-free and task-agnostic continual learning ●Approach highlight ○Prompt pool memory space allows rehearsal buffer-free and task-agnostic ○ Penalize frequently-used prompts by using prompt frequency table H_t at training time for the diversity of prompt (1) ○ At training-time, If quer..
Paper_review[short]
2023. 4. 18. 17:20