REVIEW 9 cited by
DenseFormer: Enhancing Information Flow in Transformers via Depth Weighted Averaging
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
DenseFormer: Enhancing Information Flow in Transformers via Depth Weighted Averaging
read the original abstract
The transformer architecture by Vaswani et al. (2017) is now ubiquitous across application domains, from natural language processing to speech processing and image understanding. We propose DenseFormer, a simple modification to the standard architecture that improves the perplexity of the model without increasing its size -- adding a few thousand parameters for large-scale models in the 100B parameters range. Our approach relies on an additional averaging step after each transformer block, which computes a weighted average of current and past representations -- we refer to this operation as Depth-Weighted-Average (DWA). The learned DWA weights exhibit coherent patterns of information flow, revealing the strong and structured reuse of activations from distant layers. Experiments demonstrate that DenseFormer is more data efficient, reaching the same perplexity of much deeper transformer models, and that for the same perplexity, these new models outperform transformer baselines in terms of memory efficiency and inference time.
Forward citations
Cited by 9 Pith papers
-
Delta Attention Residuals
Delta Attention Residuals attend over per-sublayer deltas instead of cumulative hidden states, producing higher-contrast attention weights and 1.7-8.2% validation perplexity gains over standard and attention residuals...
-
Do Value Vectors in Deep Layers Need Context from the Residual Stream?
Deep transformer layers can replace context-dependent value vectors with per-token lookup tables (Bank of Values), improving validation loss and the 21-benchmark average at 135M–780M while cutting FLOPs and the value cache.
-
Do Value Vectors in Deep Layers Need Context from the Residual Stream?
Deeper transformer layers benefit from context-free token-specific value vectors in a Bank of Values lookup table, improving performance over standard attention with less compute.
-
Most Transformer Modifications Still Do Not Transfer at 1-3B: A 2020-2026 Update to Narang et al. (2021) with Downstream Evaluation and a Noise Floor
Empirical update to prior work shows most of 20 recent Transformer modifications do not transfer at 1-3B scales when measured with downstream CLIMB-12 tasks, multi-seed noise floor, and cross-scale stability.
-
Multi-Head Attention Residuals
Splitting the depth-routing query in attention residuals into H per-subspace heads improves or preserves pretraining loss at 100M-1B with zero added parameters.
-
Multi-Head Attention Residuals
Splitting the depth-routing query into per-subspace heads (a parameter-free reshape) improves Transformer validation loss at 100M–1B and mid-training at 8B.
-
Hyperloop Transformers
Hyperloop Transformers outperform standard and mHC Transformers with roughly 50% fewer parameters by looping a middle block of layers and applying hyper-connections only after each loop.
-
Hyperloop Transformers
A looped Transformer with matrix-valued hyper-connections matches depth-matched baselines at ~50% fewer parameters, including under post-training quantization.
-
Attention Residuals
Attention Residuals replaces fixed residual summation with input-dependent softmax attention over preceding layers, and a blocked variant is shown to improve uniformity and downstream performance in a 48B-parameter mo...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.