REVIEW 3 cited by
The Depth-to-Width Interplay in Self-Attention
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
read the original abstract
Self-attention architectures, which are rapidly pushing the frontier in natural language processing, demonstrate a surprising depth-inefficient behavior: previous works indicate that increasing the internal representation (network width) is just as useful as increasing the number of self-attention layers (network depth). We theoretically predict a width-dependent transition between depth-efficiency and depth-inefficiency in self-attention. We conduct systematic empirical ablations on networks of depths 6 to 48 that clearly reveal the theoretically predicted behaviors, and provide explicit quantitative suggestions regarding the optimal depth-to-width allocation for a given self-attention network size. The race towards beyond 1-Trillion parameter language models renders informed guidelines for increasing self-attention depth and width in tandem an essential ingredient. Our guidelines elucidate the depth-to-width trade-off in self-attention networks of sizes up to the scale of GPT3 (which we project to be too deep for its size), and beyond, marking an unprecedented width of 30K as optimal for a 1-Trillion parameter network.
Forward citations
Cited by 3 Pith papers
-
Inverse Depth Scaling From Most Layers Being Similar
LLM loss decreases roughly inversely with depth because most layers act as a redundant ensemble that averages errors, not as a compositional hierarchy.
-
Universal pre-training by iterated random computation
Pre-training a transformer on data generated by randomly initialized LSTMs yields zero-shot in-context learning on several held-out datasets, with gains that improve with scale and faster finetuning.
-
Leaner Transformers: More Heads, Less Depth
The authors claim that more attention heads improve transformer conditioning enough to replace depth with width, yielding 30-50% parameter reductions at matched accuracy.
Discussion (0). Continue with ORCID to comment.