REVIEW 5 cited by
Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Despite the widespread success of Transformers on NLP tasks, recent works have found that they struggle to model several formal languages when compared to recurrent models. This raises the question of why Transformers perform well in practice and whether they have any properties that enable them to generalize better than recurrent models. In this work, we conduct an extensive empirical study on Boolean functions to demonstrate the following: (i) Random Transformers are relatively more biased towards functions of low sensitivity. (ii) When trained on Boolean functions, both Transformers and LSTMs prioritize learning functions of low sensitivity, with Transformers ultimately converging to functions of lower sensitivity. (iii) On sparse Boolean functions which have low sensitivity, we find that Transformers generalize near perfectly even in the presence of noisy labels whereas LSTMs overfit and achieve poor generalization accuracy. Overall, our results provide strong quantifiable evidence that suggests differences in the inductive biases of Transformers and recurrent models which may help explain Transformer's effective generalization performance despite relatively limited expressiveness.
Forward citations
Cited by 5 Pith papers
-
Can Transformers Really Do It All? On the Compatibility of Inductive Biases Across Tasks
Learned replacement non-linearities show transformers are rarely optimal for algorithmic tasks, with benefits that are task-specific, while language/code gains are smaller and more transferable.
-
When Do Neural Networks Learn World Models?
With Boolean variables, a low-degree bias, and a task distribution weighted toward simple functions of the latents, multi-task training provably recovers the latent world model up to permutations and negations.
-
Parity Requires Unified Input Dependence and Negative Eigenvalues in SSMs
For diagonal state space models, parity cannot be solved by combining input-dependent non-negative layers (Mamba) with input-independent negative-eigenvalue layers (S4D); a single layer needs both properties.
-
In-Context Learning Strategies Emerge Rationally
Transformer in-context learning is modeled as a posterior-weighted mixture of memorizing and generalizing Bayesian predictors, with a loss-complexity tradeoff governed by three fitted parameters.
-
Characterising the Inductive Biases of Neural Networks on Boolean Data
Randomly initialized depth-2 discrete networks are a priori biased toward Boolean functions with small disjunctive normal form complexity, and this bias quantitatively predicts training and generalization behavior inc...
Discussion (0). Continue with ORCID to comment.