Fangyuan Yu PRO
Ksgk-fy
AI & ML interests
AGI
Recent Activity
updated a model about 16 hours ago
Ksgk-fy/cbd-neologisms-v2 published a model about 16 hours ago
Ksgk-fy/cbd-neologisms-v2 published a model about 18 hours ago
Ksgk-fy/arl-gsm8k-drgrpo-8seedOrganizations
Representation & Optimization
Understanding about representation sheds light on optimization
-
Nuclear Norm Regularization for Deep Learning
Paper • 2405.14544 • Published • 1 -
Token embeddings violate the manifold hypothesis
Paper • 2504.01002 • Published • 1 -
Approximate Nullspace Augmented Finetuning for Robust Vision Transformers
Paper • 2403.10476 • Published • 1 -
ElaLoRA: Elastic & Learnable Low-Rank Adaptation for Efficient Model Fine-Tuning
Paper • 2504.00254 • Published • 1
Fine Dataset
-
HuggingFaceFW/fineweb
Viewer • Updated • 52.5B • 284k • 2.9k -
Reinforcement Learning-Based Prompt Template Stealing for Text-to-Image Models
Paper • 2510.00046 • Published -
2.5 Years in Class: A Multimodal Textbook for Vision-Language Pretraining
Paper • 2501.00958 • Published • 111 -
Language of Thought Shapes Output Diversity in Large Language Models
Paper • 2601.11227 • Published • 10
Embodied AI
Embodiment enables interaction of model with environment. Key is to anticipate what change could've come with its current action.
-
Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation
Paper • 2408.11812 • Published • 6 -
WebArena: A Realistic Web Environment for Building Autonomous Agents
Paper • 2307.13854 • Published • 27 -
Agent Workflow Memory
Paper • 2409.07429 • Published • 32 -
Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale
Paper • 2409.08264 • Published • 48
What I don't understand
Gap in knowledge is the place I learn the most from.
-
Recurrent Neural Networks Learn to Store and Generate Sequences using Non-Linear Representations
Paper • 2408.10920 • Published • 1 -
Towards Exact Computation of Inductive Bias
Paper • 2406.15941 • Published -
DrawingSpinUp: 3D Animation from Single Character Drawings
Paper • 2409.08615 • Published • 19 -
Learning to Move Like Professional Counter-Strike Players
Paper • 2408.13934 • Published • 23
RL
-
Accelerating Exploration with Unlabeled Prior Data
Paper • 2311.05067 • Published • 2 -
Efficient Online Reinforcement Learning with Offline Data
Paper • 2302.02948 • Published • 2 -
Learning to (Learn at Test Time): RNNs with Expressive Hidden States
Paper • 2407.04620 • Published • 34 -
Reinforcement Pre-Training
Paper • 2506.08007 • Published • 265
Exciting Papers
Our curated list of AI papers @Temus AI
-
Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models
Paper • 2310.04406 • Published • 10 -
Chain-of-Thought Reasoning Without Prompting
Paper • 2402.10200 • Published • 111 -
ICDPO: Effectively Borrowing Alignment Capability of Others via In-context Direct Preference Optimization
Paper • 2402.09320 • Published • 6 -
Self-Discover: Large Language Models Self-Compose Reasoning Structures
Paper • 2402.03620 • Published • 117
Cognition
Perception and abstraction. Each modality is tokenized and embedded into vectors for model to comprehend.
-
VILA^2: VILA Augmented VILA
Paper • 2407.17453 • Published • 41 -
Octopus v4: Graph of language models
Paper • 2404.19296 • Published • 118 -
Octo-planner: On-device Language Model for Planner-Action Agents
Paper • 2406.18082 • Published • 48 -
Dolphin: Long Context as a New Modality for Energy-Efficient On-Device Language Models
Paper • 2408.15518 • Published • 42
Memory
Prompt is text-based memory. System II prompting is updating memory. Parametric memory is long-term, while prompt-based are short-tem.
-
What You Say = What You Want? Teaching Humans to Articulate Requirements for LLMs
Paper • 2409.08775 • Published -
OmniQuery: Contextually Augmenting Captured Multimodal Memory to Enable Personal Question Answering
Paper • 2409.08250 • Published • 1 -
Synthetic continued pretraining
Paper • 2409.07431 • Published • 5 -
WonderWorld: Interactive 3D Scene Generation from a Single Image
Paper • 2406.09394 • Published • 3
Emergence
RL
-
Accelerating Exploration with Unlabeled Prior Data
Paper • 2311.05067 • Published • 2 -
Efficient Online Reinforcement Learning with Offline Data
Paper • 2302.02948 • Published • 2 -
Learning to (Learn at Test Time): RNNs with Expressive Hidden States
Paper • 2407.04620 • Published • 34 -
Reinforcement Pre-Training
Paper • 2506.08007 • Published • 265
Representation & Optimization
Understanding about representation sheds light on optimization
-
Nuclear Norm Regularization for Deep Learning
Paper • 2405.14544 • Published • 1 -
Token embeddings violate the manifold hypothesis
Paper • 2504.01002 • Published • 1 -
Approximate Nullspace Augmented Finetuning for Robust Vision Transformers
Paper • 2403.10476 • Published • 1 -
ElaLoRA: Elastic & Learnable Low-Rank Adaptation for Efficient Model Fine-Tuning
Paper • 2504.00254 • Published • 1
Exciting Papers
Our curated list of AI papers @Temus AI
-
Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models
Paper • 2310.04406 • Published • 10 -
Chain-of-Thought Reasoning Without Prompting
Paper • 2402.10200 • Published • 111 -
ICDPO: Effectively Borrowing Alignment Capability of Others via In-context Direct Preference Optimization
Paper • 2402.09320 • Published • 6 -
Self-Discover: Large Language Models Self-Compose Reasoning Structures
Paper • 2402.03620 • Published • 117
Fine Dataset
-
HuggingFaceFW/fineweb
Viewer • Updated • 52.5B • 284k • 2.9k -
Reinforcement Learning-Based Prompt Template Stealing for Text-to-Image Models
Paper • 2510.00046 • Published -
2.5 Years in Class: A Multimodal Textbook for Vision-Language Pretraining
Paper • 2501.00958 • Published • 111 -
Language of Thought Shapes Output Diversity in Large Language Models
Paper • 2601.11227 • Published • 10
Cognition
Perception and abstraction. Each modality is tokenized and embedded into vectors for model to comprehend.
-
VILA^2: VILA Augmented VILA
Paper • 2407.17453 • Published • 41 -
Octopus v4: Graph of language models
Paper • 2404.19296 • Published • 118 -
Octo-planner: On-device Language Model for Planner-Action Agents
Paper • 2406.18082 • Published • 48 -
Dolphin: Long Context as a New Modality for Energy-Efficient On-Device Language Models
Paper • 2408.15518 • Published • 42
Embodied AI
Embodiment enables interaction of model with environment. Key is to anticipate what change could've come with its current action.
-
Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation
Paper • 2408.11812 • Published • 6 -
WebArena: A Realistic Web Environment for Building Autonomous Agents
Paper • 2307.13854 • Published • 27 -
Agent Workflow Memory
Paper • 2409.07429 • Published • 32 -
Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale
Paper • 2409.08264 • Published • 48
Memory
Prompt is text-based memory. System II prompting is updating memory. Parametric memory is long-term, while prompt-based are short-tem.
-
What You Say = What You Want? Teaching Humans to Articulate Requirements for LLMs
Paper • 2409.08775 • Published -
OmniQuery: Contextually Augmenting Captured Multimodal Memory to Enable Personal Question Answering
Paper • 2409.08250 • Published • 1 -
Synthetic continued pretraining
Paper • 2409.07431 • Published • 5 -
WonderWorld: Interactive 3D Scene Generation from a Single Image
Paper • 2406.09394 • Published • 3
What I don't understand
Gap in knowledge is the place I learn the most from.
-
Recurrent Neural Networks Learn to Store and Generate Sequences using Non-Linear Representations
Paper • 2408.10920 • Published • 1 -
Towards Exact Computation of Inductive Bias
Paper • 2406.15941 • Published -
DrawingSpinUp: 3D Animation from Single Character Drawings
Paper • 2409.08615 • Published • 19 -
Learning to Move Like Professional Counter-Strike Players
Paper • 2408.13934 • Published • 23