paper-with-me

Papers

Understanding Token Probability Encoding in Output Embeddings

2024-06-03 · Hakaze Cho, Yoshihiro Sakai, Kenshiro Tanaka, Mariko Kato, Naoya Inoue

In this paper, we investigate the output token probability information in the output embedding of language models. We find an approximate common log-linear encoding of output token probabilities within the output embedding vectors and empirically demonstrate that it is accurate and sparse. As a causality examination, we steer the encoding in output embedding to modify the output probability distribution accurately. Moreover, the sparsity we find in output probability encoding suggests that a large number of dimensions in the output embedding do not contribute to causal language modeling. Therefore, we attempt to delete the output-unrelated dimensions and find more than 30% of the dimensions can be deleted without significant movement in output distribution and sequence generation. Additionally, in the pre-training dynamics of language models, we find that the output embeddings capture the corpus token frequency information in early steps, even before an obvious convergence of parameters starts.

📄 PDF Abstract BibTeX arXiv:2406.01468

Code (0)

등록된 구현이 없습니다.

Tasks

Causal Language ModelingLanguage ModelingLanguage Modelling

Similar Papers 제목 키워드 기반

Language Models over Canonical Byte-Pair Encodings

2025-06-09 · Tim Vieira, Tianyu Liu, Clemente Pasti, Yahya Emara 외

Modern language models represent probability distributions over character strings as distributions over (shorter) token strings derived via a deterministic tokenizer, such as byte-pair encoding. While this approach is hi…

valid

Do Llamas Work in English? On the Latent Language of Multilingual Transformers

2024-02-16 · Chris Wendler, Veniamin Veselovsky, Giovanni Monea, Robert West

We ask whether multilingual language models trained on unbalanced, English-dominated corpora use English as an internal pivot language -- a question of key importance for understanding how language models function and th…

Phase Diagram of Vision Large Language Models Inference: A Perspective from Interaction across Image and Instruction

2024-11-01 · Houjing Wei, Yuting Shi, Naoya Inoue

Vision Large Language Models (VLLMs) usually take input as a concatenation of image token embeddings and text token embeddings and conduct causal modeling. However, their internal behaviors remain underexplored, raising …

multimodal interaction

Pseudo-Masked Language Models for Unified Language Model Pre-Training

2020-01-01 · ICML 2020 1 · Hangbo Bao, Li Dong, Furu Wei, Wenhui Wang 외

We propose to pre-train a unified language model for both autoencoding and partially autoregressive language modeling tasks using a novel training procedure, referred to as a pseudo-masked language model (PMLM). Given an…

DecoderLanguage ModelingLanguage ModellingNatural Language Understanding+1

UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training

2020-02-28 · Hangbo Bao, Li Dong, Furu Wei, Wenhui Wang 외

We propose to pre-train a unified language model for both autoencoding and partially autoregressive language modeling tasks using a novel training procedure, referred to as a pseudo-masked language model (PMLM). Given an…

Abstractive Text SummarizationDecoderLanguage ModelingLanguage Modelling+3