dls-2023 | Deep Learning JP - Part 6

Deep Transformers without Shortcuts: Modifying Self-attention for Faithful Signal Propagation

マルチモーダル基盤モデル

TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models

HyperDiffusion: Generating Implicit Neural Fields withWeight-Space Diffusion

大量API・ツールの扱いに特化したLLM

DINOv2: Learning Robust Visual Features without Supervision

Poisoning Language Models During Instruction Tuning Instruction Tuningを利⽤した⾔語モデルのポイズニング

Egocentric Video Task Translation (CVPR 2023 Highlight)

Flow Matching for Generative Modeling

Visual Classification via Description from Large Language Models (ICLR2023)

6 / 10« 先頭 «...4 567 8...»最後 »