Log In Sign Up

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

by   Pedro A. Ortega, et al.

The recent phenomenal success of language models has reinvigorated machine learning research, and large sequence models such as transformers are being applied to a variety of domains. One important problem class that has remained relatively elusive however is purposeful adaptive behavior. Currently there is a common perception that sequence models "lack the understanding of the cause and effect of their actions" leading them to draw incorrect inferences due to auto-suggestive delusions. In this report we explain where this mismatch originates, and show that it can be resolved by treating actions as causal interventions. Finally, we show that in supervised learning, one can teach a system to condition or intervene on data by training with factual and counterfactual error signals respectively.


page 1

page 2

page 3

page 4


Counterfactual Instances Explain Little

In many applications, it is important to be able to explain the decision...

Counterfactual Temporal Point Processes

Machine learning models based on temporal point processes are the state ...

Causal Decision Making and Causal Effect Estimation Are Not the Same... and Why It Matters

Causal decision making (CDM) at scale has become a routine part of busin...

Causal Confusion in Imitation Learning

Behavioral cloning reduces policy learning to supervised learning by tra...

What if This Modified That? Syntactic Interventions via Counterfactual Embeddings

Neural language models exhibit impressive performance on a variety of ta...

Learning to Sketch with Deep Q Networks and Demonstrated Strokes

Doodling is a useful and common intelligent skill that people can learn ...

Efficient Transformers with Dynamic Token Pooling

Transformers achieve unrivalled performance in modelling language, but r...