68 reports across 10 topics

Browse machine learning by topic

Every report is filed under one topic hub, so you can start from a broad area and walk down to the specific papers that shaped it.

Fig. 02 / Field notesMeaning has a neighborhood
An illustrative embedding space: related vectors form neighborhoods, with nearest neighbors linked.
Related vectors lie near one another in an illustrative embedding space.
14 reports

Computer vision

Image classification, detection, segmentation, representation learning and neural scene representations.

  • computer-vision
  • cnn
  • object-detection
12 reports

Architectures

The network designs that mattered: attention, convolutions, residual and graph models, sequence-to-sequence and their descendants.

  • transformers
  • nlp
  • pretraining
9 reports

Applications & prompting

Reasoning, retrieval and prompting: the techniques that turn a base model into a usable system.

  • reasoning
  • nlp
  • prompting
7 reports

Efficiency & deployment

Doing more with less compute: adapters, quantisation, fast attention, sparse experts and deployment trade-offs.

  • efficiency
  • mixture-of-experts
  • quantization
6 reports

Training & data

How models are actually trained — objectives, scaling budgets, regularisation and the data decisions behind the results.

  • training
  • deep-learning
  • regularization
6 reports

Generative models

Adversarial, diffusion and latent-variable models for synthesising images and other high-dimensional data.

  • diffusion
  • generative
  • generative-models
4 reports

Reinforcement learning

Agents that learn from reward — value-based methods, policy gradients and self-play.

  • reinforcement-learning
  • deep-learning
  • ppo
4 reports

Foundations & scaling

Scaling laws, compute-optimal training and the large-model results that reframed what a language model can do.

  • scaling-laws
  • large-language-models
  • training
4 reports

Alignment & safety

Turning raw predictors into assistants: human feedback, instruction tuning, preference optimisation and self-critique.

  • alignment
  • rlhf
  • human-feedback
2 reports

Interpretability

What is actually happening inside a trained network, and how we can tell.

  • interpretability
  • computer-vision
  • visualization