Bibliography:

1. ​ ‘RL’ tag

2. ​ ‘continual learning’ tag

3. ​ ‘LM tokenization’ tag

4. ​ ‘inner monologue (AI)’ tag

5. ​ ‘hidden-information game’ tag

6. ​ ‘robotics’ tag

7. ​ ‘RL scaling’ tag

8. ​ Free-Play Periods for RL Agents

9. ​ WBE and DRL: a Middle Way of imitation learning from the human brain

10. ​ Continuous Autoregressive Models with Noise Augmentation Avoid Error Accumulation

11. ​ State-space models can learn in-context by gradient descent

12. ​ Thinking LLMs: General Instruction Following with Thought Generation

13. ​ MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering

14. ​ Contextual Document Embeddings

15. ​ Generating Diverse and Reliable Features for Few-Shot Learning

16. ​ When Parts are Greater Than Sums: Individual LLM Components Can Outperform Full Models

17. ​ Probing the Decision Boundaries of In-context Learning in Large Language Models

18. ​ Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models

19. ​ Discovering Preference Optimization Algorithms with and for Large Language Models

20. ​ State Soup: In-Context Skill Learning, Retrieval and Mixing

21. ​ Attention as a Hypernetwork

22. ​ BERTs are Generative In-Context Learners

23. ​ To Believe or Not to Believe Your LLM

24. ​ Learning to grok: Emergence of in-context learning and skill composition in modular arithmetic tasks

25. ​ Auto Evol-Instruct: Automatic Instruction Evolving for Large Language Models

26. ​ A Theoretical Understanding of Self-Correction through In-context Alignment

27. ​ MLPs Learn In-Context

28. ​ Zero-Shot Tokenizer Transfer

29. ​ Position: Understanding LLMs Requires More Than Statistical Generalization

30. ​ SOPHON: Non-Fine-Tunable Learning to Restrain Task Transferability For Pre-trained Models

31. ​ Many-Shot In-Context Learning

32. ​ Foundational Challenges in Assuring Alignment and Safety of Large Language Models

33. ​ Revisiting the Equivalence of In-Context Learning and Gradient Descent: The Impact of Data Distribution

34. ​ Best Practices and Lessons Learned on Synthetic Data for Language Models

35. ​ From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples

36. ​ Mixture-of-Depths: Dynamically allocating compute in transformer-based language models

37. ​ Evolutionary Optimization of Model Merging Recipes

38. ​ How Well Can Transformers Emulate In-context Newton’s Method?

39. ​ Revisiting Dynamic Evaluation: Online Adaptation for Large Language Models

40. ​ Neural Network Parameter Diffusion

41. ​ The Matrix: A Bayesian learning model for LLMs

42. ​ Rephrasing the Web (WARP): A Recipe for Compute and Data-Efficient Language Modeling

43. ​ An Information-Theoretic Analysis of In-Context Learning

44. ​ Deep de Finetti: Recovering Topic Distributions from Large Language Models

45. ​ Generative Multimodal Models are In-Context Learners

46. ​ VILA: On Pre-training for Visual Language Models

47. ​ Evolving Reservoirs for Meta Reinforcement Learning

48. ​ The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning

49. ​ Learning few-shot imitation as cultural transmission

50. ​ In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steering

51. ​ Rephrase and Respond: Let Large Language Models Ask Better Questions for Themselves

52. ​ ProSG: Using Prompt Synthetic Gradients to Alleviate Prompt Forgetting of RNN-like Language Models

53. ​ Self-AIXI: Self-Predictive Universal AI

54. ​ HyperFields: Towards Zero-Shot Generation of NeRFs from Text

55. ​ Transformers Learn Higher-Order Optimization Methods for In-Context Learning: A Study with Linear Models

56. ​ Eureka: Human-Level Reward Design via Coding Large Language Models

57. ​ How Many Pretraining Tasks Are Needed for In-Context Learning of Linear Regression?

58. ​ Motif: Intrinsic Motivation from Artificial Intelligence Feedback

59. ​ ExpeL: LLM Agents Are Experiential Learners

60. ​ Diversifying AI: Towards Creative Chess with AlphaZero (AZ_db)

61. ​ RAVEN: In-Context Learning with Retrieval-Augmented Encoder-Decoder Language Models

62. ​ CausalLM is not optimal for in-context learning

63. ​ MetaDiff: Meta-Learning with Conditional Diffusion for Few-Shot Learning

64. ​ Self Expanding Neural Networks

65. ​ Teaching Arithmetic to Small Transformers

66. ​ One Step of Gradient Descent is Provably the Optimal In-Context Learner with One Layer of Linear Self-Attention

67. ​ Trainable Transformer in Transformer

68. ​ Supervised Pretraining Can Learn In-Context Reinforcement Learning

69. ​ Pretraining task diversity and the emergence of non-Bayesian in-context learning for regression

70. ​ Language models are weak learners

71. ​ Minigrid & Miniworld: Modular & Customizable Reinforcement Learning Environments for Goal-Oriented Tasks

72. ​ Improving Long-Horizon Imitation Through Instruction Prediction

73. ​ Schema-learning and rebinding as mechanisms of in-context learning and emergence

74. ​ RGD: Stochastic Re-weighted Gradient Descent via Distributionally Robust Optimization

75. ​ Transformers learn to implement preconditioned gradient descent for in-context learning

76. ​ Learning Transformer Programs

77. ​ Fundamental Limitations of Alignment in Large Language Models

78. ​ How well do Large Language Models perform in Arithmetic tasks?

79. ​ Larger language models do in-context learning differently

80. ​ BiLD: Big Little Transformer Decoder

81. ​ Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery

82. ​ Looped Transformers as Programmable Computers

83. ​ A Survey of Meta-Reinforcement Learning

84. ​ Human-like systematic generalization through a meta-learning neural network

85. ​ Why Can GPT Learn In-Context? Language Models Secretly Perform Gradient Descent as Meta-Optimizers

86. ​ Unnatural Instructions: Tuning Language Models with (Almost) No Human Labor

87. ​ Rethinking the Role of Scale for In-Context Learning: An Interpretability-based Case Study at 66 Billion Scale

88. ​ Transformers learn in-context by gradient descent

89. ​ FWL: Meta-Learning Fast Weight Language Models

90. ​ What learning algorithm is in-context learning? Investigations with linear models

91. ​ Self-Destructing Models: Increasing the Costs of Harmful Dual Uses of Foundation Models

92. ​ VeLO: Training Versatile Learned Optimizers by Scaling Up

93. ​ Mysteries of mode collapse § Inescapable wedding parties

94. ​ BLOOMZ/mT0: Crosslingual Generalization through Multitask Finetuning

95. ​ ProMoT: Preserving In-Context Learning ability in Large Language Model Fine-tuning

96. ​ In-context Reinforcement Learning with Algorithm Distillation

97. ​ SAP: Bidirectional Language Models Are Also Few-shot Learners

98. ​ g.pt: Learning to Learn with Generative Models of Neural Network Checkpoints

99. ​ AlexaTM 20B: Few-Shot Learning Using a Large-Scale Multilingual Seq2Seq Model

100. ​ Few-shot Adaptation Works with UnpredicTable Data

101. ​ What Can Transformers Learn In-Context? A Case Study of Simple Function Classes

102. ​ Transformer Neural Processes: Uncertainty-Aware Meta Learning Via Sequence Modeling

103. ​ TabPFN: Meta-Learning a Real-Time Tabular AutoML Method For Small Data

104. ​ Offline RL Policies Should be Trained to be Adaptive

105. ​ Goal-Conditioned Generators of Deep Policies

106. ​ Prompting Decision Transformer for Few-Shot Policy Generalization

107. ​ RHO-LOSS: Prioritized Training on Points that are Learnable, Worth Learning, and Not Yet Learnt

108. ​ NOAH: Neural Prompt Search

109. ​ Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions

110. ​ Towards Learning Universal Hyperparameter Optimizers with Transformers

111. ​ Instruction Induction: From Few Examples to Natural Language Task Descriptions

112. ​ Gato: A Generalist Agent

113. ​ Unifying Language Learning Paradigms

114. ​ Data Distributional Properties Drive Emergent Few-Shot Learning in Transformers

115. ​ Tk-Instruct: Benchmarking Generalization via In-Context Instructions on 1,600+ Language Tasks

116. ​ What Language Model Architecture and Pretraining Objective Work Best for Zero-Shot Generalization?

117. ​ Effective Mutation Rate Adaptation through Group Elite Selection

118. ​ Searching for Efficient Neural Architectures for On-Device ML on Edge TPUs

119. ​ Can language models learn from explanations in context?

120. ​ Auto-Lambda: Disentangling Dynamic Task Relationships

121. ​ In-Context Learning and Induction Heads

122. ​ HyperMixer: An MLP-based Low Cost Alternative to Transformers

123. ​ LiteTransformerSearch: Training-free Neural Architecture Search for Efficient Language Models

124. ​ Evolving Curricula with Regret-Based Environment Design

125. ​ HyperPrompt: Prompt-based Task-Conditioning of Transformers

126. ​ Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?

127. ​ All You Need Is Supervised Learning: From Imitation Learning to Meta-RL With Upside Down RL

128. ​ NeuPL: Neural Population Learning

129. ​ Learning Synthetic Environments and Reward Networks for Reinforcement Learning

130. ​ Datamodels: Predicting Predictions from Training Data

131. ​ From data to functa: Your data point is a function and you should treat it like one

132. ​ Environment Generation for Zero-Shot Compositional Reinforcement Learning

133. ​ Active Predictive Coding Networks: A Neural Solution to the Problem of Learning Reference Frames and Part-Whole Hierarchies

134. ​ Learning robust perceptive locomotion for quadrupedal robots in the wild

135. ​ Automated Reinforcement Learning (AutoRL): A Survey and Open Problems

136. ​ In Defense of the Unitary Scalarization for Deep Multi-Task Learning

137. ​ HyperTransformer: Model Generation for Supervised and Semi-Supervised Few-Shot Learning

138. ​ Finding General Equilibria in Many-Agent Economic Simulations Using Deep Reinforcement Learning

139. ​ The Curse of Zero Task Diversity: On the Failure of Transfer Learning to Outperform MAML and their Empirical Equivalence

140. ​ A Mathematical Framework for Transformer Circuits

141. ​ PFNs: Transformers Can Do Bayesian Inference

142. ​ How to Learn and Represent Abstractions: An Investigation using Symbolic Alchemy

143. ​ Noether Networks: Meta-Learning Useful Conserved Quantities

144. ​ A rational reinterpretation of dual-process theories

145. ​ A General Language Assistant as a Laboratory for Alignment

146. ​ A Modern Self-Referential Weight Matrix That Learns to Modify Itself

147. ​ A Survey of Generalization in Deep Reinforcement Learning

148. ​ Gradients are Not All You Need

149. ​ An Explanation of In-context Learning as Implicit Bayesian Inference

150. ​ Procedural Generalization by Planning with Self-Supervised World Models

151. ​ MetaICL: Learning to Learn In Context

152. ​ Logical Activation Functions: Logit-space equivalents of Probabilistic Boolean Operators

153. ​ Shaking the foundations: delusions in sequence models for interaction and control

154. ​ Meta-learning, social cognition and consciousness in brains and machines

155. ​ T0: Multitask Prompted Training Enables Zero-Shot Task Generalization

156. ​ Replay-Guided Adversarial Environment Design

157. ​ Embodied intelligence via learning and evolution

158. ​ Transformers are Meta-Reinforcement Learners

159. ​ Scalable Online Planning via Reinforcement Learning Fine-Tuning

160. ​ Dropout’s Dream Land: Generalization from Learned Simulators to Reality

161. ​ Is Curiosity All You Need? On the Utility of Emergent Behaviors from Curious Exploration

162. ​ Bootstrapped Meta-Learning

163. ​ The Sensory Neuron as a Transformer: Permutation-Invariant Neural Networks for Reinforcement Learning

164. ​ FLAN: Finetuned Language Models Are Zero-Shot Learners

165. ​ The AI Economist: Optimal Economic Policy Design via Two-level Deep Reinforcement Learning

166. ​ Open-Ended Learning Leads to Generally Capable Agents

167. ​ Dataset Distillation with Infinitely Wide Convolutional Networks

168. ​ Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability

169. ​ PonderNet: Learning to Ponder

170. ​ Multimodal Few-Shot Learning with Frozen Language Models

171. ​ LHOPT: A Generalizable Approach to Learning Optimizers

172. ​ Towards mental time travel: a hierarchical memory for reinforcement learning agents

173. ​ A Full-stack Accelerator Search Technique for Vision Applications

174. ​ Reward is enough

175. ​ Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020

176. ​ CrossFit: A Few-shot Learning Challenge for Cross-task Generalization in NLP

177. ​ Podracer architectures for scalable Reinforcement Learning

178. ​ BLUR: Meta-Learning Bidirectional Update Rules

179. ​ Asymmetric self-play for automatic goal discovery in robotic manipulation

180. ​ OmniNet: Omnidirectional Representations from Transformers

181. ​ Linear Transformers Are Secretly Fast Weight Programmers

182. ​ Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm

183. ​ ES-ENAS: Blackbox Optimization over Hybrid Spaces via Combinatorial and Continuous Evolution

184. ​ Training Learned Optimizers with Randomly Initialized Learned Optimizers

185. ​ Evolving Reinforcement Learning Algorithms

186. ​ Meta Pseudo Labels

187. ​ Meta Learning Backpropagation And Improving It

188. ​ Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design

189. ​ Scaling down Deep Learning

190. ​ Reverse engineering learned optimizers reveals known and novel mechanisms

191. ​ Dataset Meta-Learning from Kernel Ridge-Regression

192. ​ MELD: Meta-Reinforcement Learning from Images via Latent State Models

193. ​ Meta-trained agents implement Bayes-optimal agents

194. ​ Learning not to learn: Nature versus nurture in silico

195. ​ Prioritized Level Replay

196. ​ Tasks, stability, architecture, and compute: Training more effective learned optimizers, and using them to train themselves

197. ​ Hidden Incentives for Auto-Induced Distributional Shift

198. ​ Grounded Language Learning Fast and Slow

199. ​ Matt Botvinick on the spontaneous emergence of learning algorithms

200. ​ Discovering Reinforcement Learning Algorithms

201. ​ Deep Reinforcement Learning and Its Neuroscientific Implications

202. ​ Decentralized Reinforcement Learning: Global Decision-Making via Local Economic Transactions

203. ​ Rapid Task-Solving in Novel Environments

204. ​ FBNetV3: Joint Architecture-Recipe Search using Predictor Pretraining

205. ​ GPT-3: Language Models are Few-Shot Learners

206. ​ Synthetic Petri Dish: A Novel Surrogate Model for Rapid Architecture Search

207. ​ Automatic Discovery of Interpretable Planning Strategies

208. ​ Meta-Reinforcement Learning for Robotic Industrial Insertion Tasks

209. ​ A Comparison of Methods for Treatment Assignment with an Application to Playlist Generation

210. ​ Approximate exploitability: Learning a best response in large games

211. ​ Meta-Learning in Neural Networks: A Survey

212. ​ Agent57: Outperforming the Atari Human Benchmark

213. ​ Designing Network Design Spaces

214. ​ Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutions

215. ​ Accelerating and Improving AlphaZero Using Population Based Training

216. ​ Meta-learning curiosity algorithms

217. ​ AutoML-Zero: Evolving Machine Learning Algorithms From Scratch

218. ​ AutoML-Zero: Open source code for the paper: "AutoML-Zero: Evolving Machine Learning Algorithms From Scratch"

219. ​ Effective Diversity in Population Based Reinforcement Learning

220. ​ AI Helps Warehouse Robots Pick Up New Tricks: Backed by machine learning luminaries, Covariant.ai’s bots can handle jobs previously needing a human touch

221. ​ Smooth markets: A basic mechanism for organizing gradient-based learners

222. ​ AutoML-Zero: Evolving Code That Learns

223. ​ Learning Neural Activations

224. ​ Meta-Learning without Memorization

225. ​ MetaFun: Meta-Learning with Iterative Functional Updates

226. ​ Leveraging Procedural Generation to Benchmark Reinforcement Learning

227. ​ Procgen Benchmark: We’re releasing Procgen Benchmark, 16 simple-to-use procedurally-generated environments which provide a direct measure of how quickly a reinforcement learning agent learns generalizable skills

228. ​ Increasing Generality in Machine Learning through Procedural Content Generation

229. ​ SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learning

230. ​ Optimizing Millions of Hyperparameters by Implicit Differentiation

231. ​ Learning to Predict Without Looking Ahead: World Models Without Forward Prediction

232. ​ Learning to Predict Without Looking Ahead: World Models Without Forward Prediction [blog]

233. ​ Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning

234. ​ Solving Rubik’s Cube with a Robot Hand

235. ​ Solving Rubik’s Cube with a Robot Hand [blog]

236. ​ Gradient Descent: The Ultimate Optimizer

237. ​ Data Valuation using Reinforcement Learning

238. ​ Multiplicative Interactions and Where to Find Them

239. ​ ANIL: Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML

240. ​ Emergent Tool Use From Multi-Agent Autocurricula

241. ​ Meta-Learning with Implicit Gradients

242. ​ A critique of pure learning and what artificial neural networks can learn from animal brains

243. ​ AutoML: A Survey of the State-of-the-Art

244. ​ Metalearned Neural Memory

245. ​ Algorithms for Hyper-Parameter Optimization

246. ​ Evolving the Hearthstone Meta

247. ​ Meta Reinforcement Learning

248. ​ One Epoch Is All You Need

249. ​ Compositional generalization through meta sequence-to-sequence learning

250. ​ Risks from Learned Optimization in Advanced Machine Learning Systems

251. ​ ICML 2019 Notes

252. ​ SpArSe: Sparse Architecture Search for CNNs on Resource-Constrained Microcontrollers

253. ​ AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence

254. ​ Alpha MAML: Adaptive Model-Agnostic Meta-Learning

255. ​ Reinforcement Learning, Fast and Slow

256. ​ Meta reinforcement learning as task inference

257. ​ Learning Loss for Active Learning

258. ​ Meta-learning of Sequential Strategies

259. ​ Searching for MobileNetV3

260. ​ Meta-learners’ learning dynamics are unlike learners’

261. ​ Ray Interference: a Source of Plateaus in Deep Reinforcement Learning

262. ​ AlphaX: eXploring Neural Architectures with Deep Neural Networks and Monte Carlo Tree Search

263. ​ Efficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context Variables

264. ​ Task2Vec: Task Embedding for Meta-Learning

265. ​ The Omniglot challenge: a 3-year progress report

266. ​ FIGR: Few-shot Image Generation with Reptile

267. ​ Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions

268. ​ Meta-Learning Neural Bloom Filters

269. ​ ffc205d2c6c2a415641f822e0c909bd847352e99.pdf

270. ​ Malthusian Reinforcement Learning

271. ​ Quantifying Generalization in Reinforcement Learning

272. ​ An Introduction to Deep Reinforcement Learning

273. ​ Meta-Learning: Learning to Learn Fast

274. ​ Evolving Space-Time Neural Architectures for Videos

275. ​ Understanding and correcting pathologies in the training of learned optimizers

276. ​ BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning

277. ​ Deep Reinforcement Learning

278. ​ Searching for Efficient Multi-Scale Architectures for Dense Image Prediction

279. ​ Backprop Evolution

280. ​ Learning Dexterous In-Hand Manipulation

281. ​ LEO: Meta-Learning with Latent Embedding Optimization

282. ​ Automatically Composing Representation Transformations as a Means for Generalization

283. ​ Human-level performance in first-person multiplayer games with population-based deep reinforcement learning

284. ​ Guided evolutionary strategies: Augmenting random search with surrogate gradients

285. ​ RUDDER: Return Decomposition for Delayed Rewards

286. ​ Meta-Learning Transferable Active Learning Policies by Deep Reinforcement Learning

287. ​ Fingerprint Policy Optimization for Robust Reinforcement Learning

288. ​ AutoAugment: Learning Augmentation Policies from Data

289. ​ Meta-Gradient Reinforcement Learning

290. ​ Continuous Learning in a Hierarchical Multiscale Neural Network

291. ​ Prefrontal cortex as a meta-reinforcement learning system

292. ​ Meta-Learning Update Rules for Unsupervised Representation Learning

293. ​ Reviving and Improving Recurrent Back-Propagation

294. ​ Kickstarting Deep Reinforcement Learning

295. ​ Reptile: On First-Order Meta-Learning Algorithms

296. ​ Some Considerations on Learning to Explore via Meta-Reinforcement Learning

297. ​ One Big Net For Everything

298. ​ Machine Theory of Mind

299. ​ Evolved Policy Gradients

300. ​ One-Shot Imitation from Observing Humans via Domain-Adaptive Meta-Learning

301. ​ Rover Descent: Learning to optimize by learning to navigate on prototypical loss surfaces

302. ​ ScreenerNet: Learning Self-Paced Curriculum for Deep Neural Networks

303. ​ Population Based Training of Neural Networks

304. ​ BlockDrop: Dynamic Inference Paths in Residual Networks

305. ​ Learning to select computations

306. ​ Learning to Generalize: Meta-Learning for Domain Generalization

307. ​ Efficient K-shot Learning with Regularized Deep Networks

308. ​ Online Learning of a Memory for Learning Rates

309. ​ One-Shot Visual Imitation Learning via Meta-Learning

310. ​ Supervising Unsupervised Learning

311. ​ Learning with Opponent-Learning Awareness

312. ​ SMASH: One-Shot Model Architecture Search through HyperNetworks

313. ​ Stochastic Optimization with Bandit Sampling

314. ​ A Simple Neural Attentive Meta-Learner

315. ​ Reinforcement Learning for Learning Rate Control

316. ​ Metacontrol for Adaptive Imagination-Based Optimization

317. ​ Deciding How to Decide: Dynamic Routing in Artificial Neural Networks

318. ​ Prototypical Networks for Few-shot Learning

319. ​ Learned Optimizers that Scale and Generalize

320. ​ MAML: Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

321. ​ Learning to Optimize Neural Nets

322. ​ Understanding Synthetic Gradients and Decoupled Neural Interfaces

323. ​ Optimization as a Model for Few-Shot Learning

324. ​ Learning to superoptimize programs

325. ​ Discovering objects and their relations from entangled scene representations

326. ​ Google Vizier: A Service for Black-Box Optimization

327. ​ An Actor-critic Algorithm for Learning Rate Learning

328. ​ A Bird’s Eye View of Synthetic Gradients

329. ​ Learning to reinforcement learn

330. ​ Learning to Learn without Gradient Descent by Gradient Descent

331. ​ RL²: Fast Reinforcement Learning via Slow Reinforcement Learning

332. ​ Designing Neural Network Architectures using Reinforcement Learning

333. ​ Using Fast Weights to Attend to the Recent Past

334. ​ HyperNetworks

335. ​ Decoupled Neural Interfaces using Synthetic Gradients

336. ​ Learning to learn by gradient descent by gradient descent

337. ​ Matching Networks for One Shot Learning

338. ​ Learning to Optimize

339. ​ One-shot Learning with Memory-Augmented Neural Networks

340. ​ Adaptive Computation Time for Recurrent Neural Networks

341. ​ On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models

342. ​ Gradient-based Hyperparameter Optimization through Reversible Learning

343. ​ Machine Teaching: an Inverse Problem to Machine Learning and an Approach Toward Optimal Education

344. ​ Human-Level Concept Learning through Probabilistic Program Induction

345. ​ abf7f07b81284b90744432cca3bf0d8cee85e469.pdf

346. ​ Robots that can adapt like animals

347. ​ Deep Learning in Neural Networks: An Overview

348. ​ Practical Bayesian Optimization of Machine Learning Algorithms

349. ​ Optimal Ordered Problem Solver (OOPS)

350. ​ Learning to Learn Using Gradient Descent

351. ​ On the Optimization of a Synaptic Learning Rule

352. ​ Interactions between Learning and Evolution

353. ​ Learning to Control Fast-Weight Memories: An Alternative to Dynamic Recurrent Networks

354. ​ Learning a synaptic learning rule

355. ​ Reinforcement Learning: An Introduction § Designing Reward Signals

356. ​ 2ddcafc570cef087ed62b0113ee2917df3a4f33a.pdf#page=491

357. ​ Exploring Hyperparameter Meta-Loss Landscapes With Jax

358. ​ f8c247b7a53735e17642638a18b45a312a3cf84f.html#google

359. ​ Metalearning

360. ​ Universal Search § OOPS and Other Incremental Variations

361. ​ Extrapolating to Unnatural Language Processing With GPT-3’s In-Context Learning: The Good, the Bad, and the Mysterious

362. ​ 53906a0a199a213fa1bce0b97ecad6b5063931e4.html

363. ​ How Does In-Context Learning Work? A Framework for Understanding the Differences from Traditional Supervised Learning

364. ​ bdf17c80e1ed5dc516811f03acef03415b220143.html

365. ​ Rapid Motor Adaptation for Legged Robots

366. ​ Collaborating With Humans Requires Understanding Them

367. ​ Why Generalization in RL Is Difficult: Epistemic POMDPs and Implicit Partial Observability [Blog]

368. ​ Hypernetworks [Blog]

369. ​ Action and Perception As Divergence Minimization

370. ​ AlphaStar: Mastering the Real-Time Strategy Game StarCraft II

371. ​ Prefrontal Cortex As a Meta-Reinforcement Learning System [Blog]

372. ​ c78ec48980d210d2a589b077f45a8d1f9e303dce.html

373. ​ The Lie Comes First, the Worlds to Accommodate It

374. ​ Sgdstore/experiments/omniglot at Master

375. ​ 864f04985a5c15c527c5b4465cc42d8707e7bbb5.html#omniglot

376. ​ Curriculum For Reinforcement Learning

377. ​ 82f88b3c03ad0b252a33823c09b770832a95bbcc.html#openai

378. ​ Neural Architecture Search

379. ​ e6974ff95b5a7295d018b89e865e011d9f66369b.html#openai

380. ​ MetaGenRL: Improving Generalization in Meta Reinforcement Learning

381. ​ cdfbfecb16cd0249c87142b4c650a487782f1399.html

382. ​ 2022: 25-Year Anniversary: LSTM (1997), All Computable Metaverses, Hierarchical Q-Learning, Adversarial Intrinsic Reinforcement Learning, Low-Complexity NNs, Low-Complexity Art, Meta-RL, Soccer Learning

383. ​ Metalearning or Learning to Learn Since 1987

384. ​ 76c24cf0db4abf4ce2b77d22182272d8e62d1a28.html

385. ​ The Future of Artificial Intelligence Is Self-Organizing and Self-Assembling

386. ​ Domain-Adaptive Meta-Learning

387. ​ 31439931d05db1bcf5dd08d9e9a09c6be5b970a0.html

388. ​ How to Fix Reinforcement Learning

389. ​ Introducing Adept

390. ​ Optimal Learning: Computational Procedures for Bayes-Adaptive Markov Decision Processes

391. ​ Risks from Learned Optimization: Introduction

392. ​ How Good Are LLMs at Doing ML on an Unknown Dataset?

393. ​ Early Situational Awareness and Its Implications, a Story

394. ​ 20ea9a879c0915ecfa2f2f87dba168dc160967cb.html

395. ​ AI Is Learning How to Create Itself

396. ​ 2874fc1c58ced20eb47a85ca46af67972fcf2ab4.html

397. ​ Matt Botvinick: Neuroscience, Psychology, and AI at DeepMind

398. ​ SMASH: One-Shot Model Architecture Search through HyperNetworks [Video]

399. ​ Solving Rubik’s Cube With a Robot Hand: Perturbations

400. ​ WELM

401. ​ 2023-lee-figure9-arithmeticcanbelearnedevenwithnoiseintheinnermonologuetranscripts.jpg

402. ​ 2022-patel-figure1-mt5fewshotpromptingwordbywordforneuralmachinetranslation.png

403. ​ 2020-real-googlebrain-automlzero-bestalgorithmannotation.mp4

404. ​ 2018-metz-appendix-figure1-detailedschematicdiagramofmetalearningarchitecture.png

405. ​ 2018-metz-figure1-schematicofmetalearningrepresentationsforunsupervisedlearning.jpg

406. ​ 2018-metz-figure5-generalizationofmetalearnedruletounseenlayesrunitsandactivations.jpg

407. ​ 2018-metz-figure6-learnedfiltersandrepresentationsofthemetalearnednet.jpg

408. ​ https://ai.facebook.com/blog/harmful-content-can-evolve-quickly-our-new-ai-system-adapts-to-tackle-it

409. ​ https://ai.meta.com/blog/harmful-content-can-evolve-quickly-our-new-ai-system-adapts-to-tackle-it/

410. ​ https://blog.waymo.com/2020/04/using-automated-data-augmentation-to.html#google

411. ​ 15822ba1a05b1de73ae8418eed6271516139faf5.html#google

412. ​ https://openai.com/blog/reptile/

413. ​ https://openai.com/index/mle-bench/

414. ​ https://pages.ucsd.edu/~rbelew/courses/cogs184_w10/readings/HintonNowlan97.pdf

415. ​ eb17a655841741a210b6b187477db8d4b5106390.pdf

416. ​ https://plato.stanford.edu/entries/selection-units/

417. ​ https://research.google/blog/introducing-flan-more-generalizable-language-models-with-instruction-fine-tuning/

418. ​ https://research.google/blog/permutation-invariant-neural-networks-for-reinforcement-learning/

419. ​ https://research.google/blog/training-machine-learning-models-more-efficiently-with-dataset-distillation/

420. ​ https://www.alignmentforum.org/posts/YEioD8YLgxih3ydxP/why-simulator-ais-want-to-be-active-inference-ais

421. ​ https://www.deepmind.com/research/publications/alchemy

422. ​ https://www.lesswrong.com/posts/QNQuWB3hS5FrGp5yZ/programmatic-backdoors-dnns-can-use-sgd-to-run-arbitrary

423. ​ https://www.lesswrong.com/posts/Y4rrwkopoigaNGxmS/a-mind-needn-t-be-curious-to-reap-the-benefits-of-curiosity#XeAmDn3NsMqdF6Mij

424. ​ https://www.lesswrong.com/posts/bC5xd7wQCnTDw7Kyx/getting-up-to-speed-on-the-speed-prior-in-2022

425. ​ https://www.lesswrong.com/posts/ddR8dExcEFJKJtWvR/how-evolutionary-lineages-of-llms-can-plan-their-own-futur

426. ​ https://www.lesswrong.com/posts/sY3a4Rfa48CgteBEm/chatgpt-can-learn-indirect-control

427. ​ 149e0b85a156eeb708a4da09b8416d59170e30ac.html

428. ​ https://www.nature.com/articles/s41467-020-19244-4#deepmind

429. ​ https://www.nature.com/articles/s42256-018-0006-z#uber

430. ​ https://www.quantamagazine.org/researchers-build-ai-that-builds-ai-20220125/

431. ​ https://www.reddit.com/r/reinforcementlearning/search/?q=flair%3AMeta-RL&sort=top&restrict_sr=on&t=all

432. ​ e1043f593c393ed0947ea9af41a2009be24d6867.html

433. ​ https://x.com/SullyOmarr/status/1768744880673522083

434. ​ https://x.com/_jasonwei/status/1587858146948567041

435. ​ https://x.com/alexalbert__/status/1636488551817965568

436. ​ https://x.com/francoisfleuret/status/1714531085512544760

437. ​ https://x.com/goodside/status/1652496489241878533

438. ​ https://x.com/joshwhiton/status/1770870746010513571

439. ​ https://x.com/shinboson/status/1794570054165729303

440. ​ MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering

441. ​ Lil'Log

442. ​ Homepage: Aleksander Mądry

443. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2410.07095%2523openai.html

444. ​ When Parts are Greater Than Sums: Individual LLM Components Can Outperform Full Models

445. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2406.13131.html

446. ​ Probing the Decision Boundaries of In-context Learning in Large Language Models

447. ​ Aditya Grover

448. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2406.11233.html

449. ​ Zero-Shot Tokenizer Transfer

450. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2405.07883.html

451. ​ Revisiting the Equivalence of In-Context Learning and Gradient Descent: The Impact of Data Distribution

452. ​ https%253A%252F%252Fieeexplore.ieee.org%252Fabstract%252Fdocument%252F10446522.html

453. ​ From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples

454. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2404.07544.html

455. ​ Rephrasing the Web (WARP): A Recipe for Compute and Data-Efficient Language Modeling

456. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2401.16380%2523apple.html

457. ​ Learning few-shot imitation as cultural transmission

458. ​ https%253A%252F%252Fwww.nature.com%252Farticles%252Fs41467-023-42875-2%2523deepmind.html

459. ​ Self-AIXI: Self-Predictive Universal AI

460. ​ https%253A%252F%252Fopenreview.net%252Fforum%253Fid%253DpsXVkKO9No%2523deepmind.html

461. ​ Diversifying AI: Towards Creative Chess with AlphaZero (AZ_db)

462. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2308.09175%2523deepmind.html

463. ​ Teaching Arithmetic to Small Transformers

464. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2307.03381.html

465. ​ Supervised Pretraining Can Learn In-Context Reinforcement Learning

466. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2306.14892.html

467. ​ Minigrid & Miniworld: Modular & Customizable Reinforcement Learning Environments for Goal-Oriented Tasks

468. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2306.13831.html

469. ​ Schema-learning and rebinding as mechanisms of in-context learning and emergence

470. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2307.01201%2523deepmind.html

471. ​ RGD: Stochastic Re-weighted Gradient Descent via Distributionally Robust Optimization

472. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2306.09222%2523google.html

473. ​ How well do Large Language Models perform in Arithmetic tasks?

474. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2304.02015%2523alibaba.html

475. ​ Larger language models do in-context learning differently

476. ​ Jason Wei

477. ​ Yi Tay

478. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2303.03846%2523google.html

479. ​ Transformers learn in-context by gradient descent

480. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2212.07677%2523google.html

481. ​ FWL: Meta-Learning Fast Weight Language Models

482. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2212.02475%2523google.html

483. ​ What learning algorithm is in-context learning? Investigations with linear models

484. ​ Jacob Andreas @ MIT

485. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2211.15661%2523google.html

486. ​ BLOOMZ/mT0: Crosslingual Generalization through Multitask Finetuning

487. ​ Thomas Wang

488. ​ Stella Biderman

489. ​ Teven Le Scao

490. ​ Sheng Shen’s Homepage

491. ​ Colin Raffel

492. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2211.01786.html

493. ​ SAP: Bidirectional Language Models Are Also Few-shot Learners

494. ​ Colin Raffel

495. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2209.14500.html

496. ​ g.pt: Learning to Learn with Generative Models of Neural Network Checkpoints

497. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2209.12892.html

498. ​ AlexaTM 20B: Few-Shot Learning Using a Large-Scale Multilingual Seq2Seq Model

499. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2208.01448%2523amazon.html

500. ​ What Can Transformers Learn In-Context? A Case Study of Simple Function Classes

501. ​ Percy Liang

502. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2208.01066.html

503. ​ TabPFN: Meta-Learning a Real-Time Tabular AutoML Method For Small Data

504. ​ Profile – Machine Learning Lab

505. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2207.01848.html

506. ​ Prompting Decision Transformer for Few-Shot Policy Generalization

507. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2206.13499.html

508. ​ RHO-LOSS: Prioritized Training on Points that are Learnable, Worth Learning, and Not Yet Learnt

509. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2206.07137.html

510. ​ Towards Learning Universal Hyperparameter Optimizers with Transformers

511. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2205.13320%2523google.html

512. ​ Gato: A Generalist Agent

513. ​ Nicolas Heess

514. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2205.06175%2523deepmind.html

515. ​ Unifying Language Learning Paradigms

516. ​ Yi Tay

517. ​ Neil Houlsby

518. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2205.05131%2523google.html

519. ​ Tk-Instruct: Benchmarking Generalization via In-Context Instructions on 1,600+ Language Tasks

520. ​ Yizhong Wang—University of Washington

521. ​ Noah A. Smith

522. ​ Hannaneh Hajishirzi—University of Washington

523. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2204.07705.html

524. ​ HyperMixer: An MLP-based Low Cost Alternative to Transformers

525. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2203.03691.html

526. ​ LiteTransformerSearch: Training-free Neural Architecture Search for Efficient Language Models

527. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2203.02094%2523microsoft.html

528. ​ HyperPrompt: Prompt-based Task-Conditioning of Transformers

529. ​ Yi Tay

530. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2203.00759.html

531. ​ Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?

532. ​ Mike Lewis

533. ​ Hannaneh Hajishirzi—University of Washington

534. ​ Luke Zettlemoyer

535. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2202.12837%2523facebook.html

536. ​ NeuPL: Neural Population Learning

537. ​ Nicolas Heess

538. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2202.07415%2523deepmind.html

539. ​ Learning robust perceptive locomotion for quadrupedal robots in the wild

540. ​ %252Fdoc%252Freinforcement-learning%252Fmeta-learning%252F2022-miki.pdf.html

541. ​ PFNs: Transformers Can Do Bayesian Inference

542. ​ Profile – Machine Learning Lab

543. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2112.10510.html

544. ​ A General Language Assistant as a Laboratory for Alignment

545. ​ About Me

546. ​ Andy Jones

547. ​ https://jack-clark.net/about/

548. ​ Sam McCandlish

549. ​ Jared Kaplan

550. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2112.00861%2523anthropic.html

551. ​ Procedural Generalization by Planning with Self-Supervised World Models

552. ​ Julian Schrittwieser

553. ​ Sherjil Ozair

554. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2111.01587%2523deepmind.html

555. ​ LHOPT: A Generalizable Approach to Learning Optimizers

556. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2106.00958%2523openai.html

557. ​ Reward is enough

558. ​ https%253A%252F%252Fwww.sciencedirect.com%252Fscience%252Farticle%252Fpii%252FS0004370221000862%2523deepmind.html

559. ​ Podracer architectures for scalable Reinforcement Learning

560. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2104.06272%2523deepmind.html

561. ​ OmniNet: Omnidirectional Representations from Transformers

562. ​ Yi Tay

563. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2103.01075%2523google.html

564. ​ Meta Pseudo Labels

565. ​ Zihang Dai

566. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2003.10580%2523google.html

567. ​ Scaling down Deep Learning

568. ​ About Sam Greydanus

569. ​ https%253A%252F%252Fgreydanus.github.io%252F2020%252F12%252F01%252Fscaling-down%252F.html

570. ​ Matt Botvinick on the spontaneous emergence of learning algorithms

571. ​ https%253A%252F%252Fwww.lesswrong.com%252Fposts%252FWnqua6eQkewL3bqsF%252Fmatt-botvinick-on-the-spontaneous-emergence-of-learning.html

572. ​ Accelerating and Improving AlphaZero Using Population Based Training

573. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2003.06212.html

574. ​ Procgen Benchmark: We’re releasing Procgen Benchmark, 16 simple-to-use procedurally-generated environments which provide a direct measure of how quickly a reinforcement learning agent learns generalizable skills

575. ​ Jacob Hilton's Homepage

576. ​ John Schulman’s Homepage

577. ​ https%253A%252F%252Fopenai.com%252Fresearch%252Fprocgen-benchmark.html

578. ​ One Epoch Is All You Need

579. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1906.06669.html

580. ​ ICML 2019 Notes

581. ​ https%253A%252F%252Fdavid-abel.github.io%252Fnotes%252Ficml_2019.pdf.html

582. ​ Meta-learners’ learning dynamics are unlike learners’

583. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1905.01320%2523deepmind.html

584. ​ Ray Interference: a Source of Plateaus in Deep Reinforcement Learning

585. ​ https://sites.google.com/view/razp/home

586. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1904.11455%2523deepmind.html

587. ​ RUDDER: Return Decomposition for Delayed Rewards

588. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1806.07857.html

589. ​ AutoAugment: Learning Augmentation Policies from Data

590. ​ Barret Zoph

591. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1805.09501%2523google.html

592. ​ Meta-Learning Update Rules for Unsupervised Representation Learning

593. ​ Luke Metz

594. ​ Jascha Sohl-Dickstein

595. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1804.00222%2523google.html

596. ​ Reptile: On First-Order Meta-Learning Algorithms

597. ​ John Schulman’s Homepage

598. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1803.02999%2523openai.html

599. ​ SMASH: One-Shot Model Architecture Search through HyperNetworks

600. ​ https%253A%252F%252Farxiv.org%252Fabs%252F1708.05344.html

601. ​ Machine Teaching: an Inverse Problem to Machine Learning and an Approach Toward Optimal Education

602. ​ %252Fdoc%252Fai%252F2015-zhu-2.pdf.html

603. ​ Optimal Ordered Problem Solver (OOPS)

604. ​ https%253A%252F%252Farxiv.org%252Fabs%252Fcs%252F0207097%2523schmidhuber.html

605. ​ Learning a synaptic learning rule

606. ​ %252Fdoc%252Freinforcement-learning%252Fmeta-learning%252F1991-bengio.pdf.html

607. Wikipedia Bibliography:

     1. ​ Automated Machine Learning

     2. ​ Baldwin Effect

     3. ​ De Finetti’s Theorem

     4. ​ Gödel Machine

     5. ​ HyperNEAT

     6. ​ Jürgen Schmidhuber

     7. ​ Long Short-Term Memory

     8. ​ Meta Learning (computer Science)

     9. ​ Price Equation

     10. ​ Prior Probability

     11. ​ Quasispecies Model

     12. ​ Sufficient Statistic

     13. ​ Viral Quasispecies

     14. ​ Ivan Vulić

     15. ​ Richard Everett

     16. ​ Sukhdeep Singh

     17. ​ Marcus Hutter

     18. ​ Lisa Schut

     19. ​ Demis Hassabis

     20. ​ Satinder Singh

     21. ​ Chelsea Finn

     22. ​ Murray Shanahan

     23. ​ Dileep George

     24. ​ Geoffrey Hinton

     25. ​ Mohammad Norouzi

     26. ​ Dragomir Radev

     27. ​ Alexei A. Efros

     28. ​ Jitendra Malik

     29. ​ Rahul Gupta

     30. ​ Haidar Khan

     31. ​ Joshua Tenenbaum

     32. ​ Aidan Gomez

     33. ​ Yutian Chen

     34. ​ Zi Wang

     35. ​ Nando De Freitas

     36. ​ Alexander Novikov

     37. ​ Jackie Kay

     38. ​ Tom Eccles

     39. ​ Oriol Vinyals

     40. ​ Yejin Choi

     41. ​ Sebastien Bubeck

     42. ​ Farinaz Koushanfar

     43. ​ Zhao Chen

     44. ​ Ed H. Chi

     45. ​ Ari Holtzman

     46. ​ Mikel Artetxe

     47. ​ Vladlen Koltun

     48. ​ Danny Hernandez

     49. ​ Dario Amodei

     50. ​ Jie Tang

     51. ​ Wojciech Zaremba

     52. ​ David Silver (programmer)

     53. ​ Doina Precup

     54. ​ Richard S. Sutton

     55. ​ John Quan

     56. ​ Quoc V. Le

     57. ​ I-Chen Wu

     58. ​ Sepp Hochreiter

     59. ​ Andrew Brock

     60. ​ Juergen Schmidhuber

     61. ​ Yoshua Bengio

     62. ​ Samy Bengio