Mechanistic Interpretability of Binary and Ternary Transformers

• URL: http://arxiv.org/abs/2405.17703v1
• Date: Mon, 27 May 2024 23:22:23 GMT
• Title: Mechanistic Interpretability of Binary and Ternary Transformers
• Authors: Jason Li,
• Abstract summary: We investigate whether binary and ternary transformer networks learn distinctly different or similar algorithms when compared to full-precision transformer networks. This provides evidence against the possibility of using binary and ternary networks as a more interpretable alternative in the Large Language Models setting.
• Score: 1.3715396507106912
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent research (arXiv:2310.11453, arXiv:2402.17764) has proposed binary and ternary transformer networks as a way to significantly reduce memory and improve inference speed in Large Language Models (LLMs) while maintaining accuracy. In this work, we apply techniques from mechanistic interpretability to investigate whether such networks learn distinctly different or similar algorithms when compared to full-precision transformer networks. In particular, we reverse engineer the algorithms learned for the toy problem of modular addition where we find that binary and ternary networks learn similar algorithms as full precision networks. This provides evidence against the possibility of using binary and ternary networks as a more interpretable alternative in the LLM setting.

Related papers

• InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques [1.075853016351821]
  We train simple neural networks using a stricter version of Interchange Intervention Training (IIT) which we call Strict IIT (SIIT) We evaluate SIIT on sparse transformers produced by the Tracr tool and find that SIIT models maintain Tracr's original circuit while being more realistic.
  arXiv  Detail & Related papers  (2024-07-19T17:46:51Z)
• Transformers meet Neural Algorithmic Reasoners [16.5785372289558]
  We propose a novel approach that combines the Transformer's language understanding with the robustness of graph neural network (GNN)-based neural algorithmic reasoners (NARs) We evaluate our resulting TransNAR model on CLRS-Text, the text-based version of the CLRS-30 benchmark, and demonstrate significant gains over Transformer-only models for algorithmic reasoning.
  arXiv  Detail & Related papers  (2024-06-13T16:42:06Z)
• NAR-Former V2: Rethinking Transformer for Universal Neural Network Representation Learning [25.197394237526865]
  We propose a modified Transformer-based universal neural network representation learning model NAR-Former V2. Specifically, we take the network as a graph and design a straightforward tokenizer to encode the network into a sequence. We incorporate the inductive representation learning capability of GNN into Transformer, enabling Transformer to generalize better when encountering unseen architecture.
  arXiv  Detail & Related papers  (2023-06-19T09:11:04Z)
• Transformers as Statisticians: Provable In-Context Learning with In-Context Algorithm Selection [88.23337313766353]
  This work first provides a comprehensive statistical theory for transformers to perform ICL. We show that transformers can implement a broad class of standard machine learning algorithms in context. A emphsingle transformer can adaptively select different base ICL algorithms.
  arXiv  Detail & Related papers  (2023-06-07T17:59:31Z)
• Centered Self-Attention Layers [89.21791761168032]
  The self-attention mechanism in transformers and the message-passing mechanism in graph neural networks are repeatedly applied. We show that this application inevitably leads to oversmoothing, i.e., to similar representations at the deeper layers. We present a correction term to the aggregating operator of these mechanisms.
  arXiv  Detail & Related papers  (2023-06-02T15:19:08Z)
• Error Correction Code Transformer [92.10654749898927]
  We propose to extend for the first time the Transformer architecture to the soft decoding of linear codes at arbitrary block lengths. We encode each channel's output dimension to high dimension for better representation of the bits information to be processed separately. The proposed approach demonstrates the extreme power and flexibility of Transformers and outperforms existing state-of-the-art neural decoders by large margins at a fraction of their time complexity.
  arXiv  Detail & Related papers  (2022-03-27T15:25:58Z)
• Multitask Learning and Joint Optimization for Transformer-RNN-Transducer Speech Recognition [13.198689566654107]
  This paper explores multitask learning, joint optimization, and joint decoding methods for transformer-RNN-transducer systems. We show that the proposed methods can reduce word error rate (WER) by 16.6 % and 13.3 % for test-clean and test-other datasets, respectively.
  arXiv  Detail & Related papers  (2020-11-02T06:38:06Z)
• High-Capacity Expert Binary Networks [56.87581500474093]
  Network binarization is a promising hardware-aware direction for creating efficient deep models. Despite its memory and computational advantages, reducing the accuracy gap between binary models and their real-valued counterparts remains an unsolved challenging research problem. We propose Expert Binary Convolution, which, for the first time, tailors conditional computing to binary networks by learning to select one data-specific expert binary filter at a time conditioned on input features.
  arXiv  Detail & Related papers  (2020-10-07T17:58:10Z)
• Binarizing MobileNet via Evolution-based Searching [66.94247681870125]
  We propose a use of evolutionary search to facilitate the construction and training scheme when binarizing MobileNet. Inspired by one-shot architecture search frameworks, we manipulate the idea of group convolution to design efficient 1-Bit Convolutional Neural Networks (CNNs) Our objective is to come up with a tiny yet efficient binary neural architecture by exploring the best candidates of the group convolution.
  arXiv  Detail & Related papers  (2020-05-13T13:25:51Z)
• Exploring the Connection Between Binary and Spiking Neural Networks [1.329054857829016]
  We bridge the recent algorithmic progress in training Binary Neural Networks and Spiking Neural Networks. We show that training Spiking Neural Networks in the extreme quantization regime results in near full precision accuracies on large-scale datasets.
  arXiv  Detail & Related papers  (2020-02-24T03:46:51Z)
• Computational optimization of convolutional neural networks using separated filters architecture [69.73393478582027]
  We consider a convolutional neural network transformation that reduces computation complexity and thus speedups neural network processing. Use of convolutional neural networks (CNN) is the standard approach to image recognition despite the fact they can be too computationally demanding.
  arXiv  Detail & Related papers  (2020-02-18T17:42:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.