Emergence and Effectiveness of Task Vectors in In-Context Learning: An Encoder Decoder Perspective

• URL: http://arxiv.org/abs/2412.12276v3
• Date: Mon, 02 Jun 2025 12:55:12 GMT
• Title: Emergence and Effectiveness of Task Vectors in In-Context Learning: An Encoder Decoder Perspective
• Authors: Seungwook Han, Jinyeop Song, Jeff Gore, Pulkit Agrawal,
• Abstract summary: We study how transformers form task vectors during pretraining and how their task encoding quality predicts ICL task performance.<n>Our empirical insights shed light into better understanding the success and failure modes of large language models via their representations.
• Score: 18.077009146950473
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Autoregressive transformers exhibit adaptive learning through in-context learning (ICL), which begs the question of how. Prior works have shown that transformers represent the ICL tasks as vectors in their representations. In this paper, we leverage the encoding-decoding framework to study how transformers form task vectors during pretraining and how their task encoding quality predicts ICL task performance. On synthetic ICL tasks, we analyze the training dynamics of a small transformer and report the coupled emergence of task encoding and decoding. As the model learns to encode different latent tasks (e.g., "Finding the first noun in a sentence.") into distinct, separable representations, it concurrently builds conditional decoding algorithms and improves its ICL performance. We validate this phenomenon across pretrained models of varying scales (Gemma-2 2B/9B/27B, Llama-3.1 8B/70B) and over the course of pretraining in OLMo-7B. Further, we demonstrate that the quality of task encoding inferred from representations predicts ICL performance, and that, surprisingly, finetuning the earlier layers can improve the task encoding and performance more than finetuning the latter layers. Our empirical insights shed light into better understanding the success and failure modes of large language models via their representations.

Related papers

• Take Off the Training Wheels Progressive In-Context Learning for Effective Alignment [22.224737528266598]
  In this paper, we investigate the impact of demonstrations on token representations within alignment tasks.<n>We propose an efficient Progressive In-Context Alignment (PICA) method consisting of two stages.<n>Our work highlights the application of ICL for alignment and calls for a deeper understanding of ICL for complex generations.
  arXiv  Detail & Related papers  (2025-03-13T02:01:02Z)
• Learning Task Representations from In-Context Learning [73.72066284711462]
  Large language models (LLMs) have demonstrated remarkable proficiency in in-context learning.<n>We introduce an automated formulation for encoding task information in ICL prompts as a function of attention heads.<n>We show that our method's effectiveness stems from aligning the distribution of the last hidden state with that of an optimally performing in-context-learned model.
  arXiv  Detail & Related papers  (2025-02-08T00:16:44Z)
• Understanding Knowledge Hijack Mechanism in In-context Learning through Associative Memory [37.93644115914534]
  In-context learning (ICL) enables large language models to adapt to new tasks without fine-tuning.<n>This paper investigates the balance between in-context information and pretrained bigram knowledge in token prediction.
  arXiv  Detail & Related papers  (2024-12-16T05:33:05Z)
• A Progressive Transformer for Unifying Binary Code Embedding and Knowledge Transfer [15.689556592544667]
  We introduce ProTST, a novel transformer-based methodology for binary code embedding.<n>ProTST employs a hierarchical training process based on a unique tree-like structure.<n>Results show that ProTST yields an average validation score (F1, MRR, and Recall@1) improvement of 14.8% compared to traditional two-stage training.
  arXiv  Detail & Related papers  (2024-12-15T13:04:29Z)
• Sparse autoencoders reveal selective remapping of visual concepts during adaptation [54.82630842681845]
  Adapting foundation models for specific purposes has become a standard approach to build machine learning systems. We develop a new Sparse Autoencoder (SAE) for the CLIP vision transformer, named PatchSAE, to extract interpretable concepts.
  arXiv  Detail & Related papers  (2024-12-06T18:59:51Z)
• Provably Transformers Harness Multi-Concept Word Semantics for Efficient In-Context Learning [53.685764040547625]
  Transformer-based large language models (LLMs) have displayed remarkable creative prowess and emergence capabilities. This work provides a fine mathematical analysis to show how transformers leverage the multi-concept semantics of words to enable powerful ICL and excellent out-of-distribution ICL abilities.
  arXiv  Detail & Related papers  (2024-11-04T15:54:32Z)
• A Theoretical Perspective for Speculative Decoding Algorithm [60.79447486066416]
  One effective way to accelerate inference is emphSpeculative Decoding, which employs a small model to sample a sequence of draft tokens and a large model to validate. This paper tackles this gap by conceptualizing the decoding problem via markov chain abstraction and studying the key properties, emphoutput quality and inference acceleration, from a theoretical perspective.
  arXiv  Detail & Related papers  (2024-10-30T01:53:04Z)
• Interpreting token compositionality in LLMs: A robustness analysis [10.777646083061395]
  Constituent-Aware Pooling (CAP) is a methodology designed to analyse how large language models process linguistic structures. CAP intervenes in model activations through constituent-based pooling at various model levels.
  arXiv  Detail & Related papers  (2024-10-16T18:10:50Z)
• How Do Nonlinear Transformers Learn and Generalize in In-Context Learning? [82.51626700527837]
  Transformer-based large language models displayed impressive in-context learning capabilities, where a pre-trained model can handle new tasks without fine-tuning. We analyze how the mechanics of how Transformer to achieve ICL contribute to the technical challenges of the training problems in Transformers.
  arXiv  Detail & Related papers  (2024-02-23T21:07:20Z)
• A Thorough Examination of Decoding Methods in the Era of LLMs [72.65956436513241]
  Decoding methods play an indispensable role in converting language models from next-token predictors into practical task solvers. This paper provides a comprehensive and multifaceted analysis of various decoding methods within the context of large language models. Our findings reveal that decoding method performance is notably task-dependent and influenced by factors such as alignment, model size, and quantization.
  arXiv  Detail & Related papers  (2024-02-10T11:14:53Z)
• Code Representation Learning At Scale [75.04686476303436]
  We fuel code representation learning with a vast amount of code data via a two-stage pretraining scheme. We first train the encoders via a mix that leverages both randomness in masking language modeling and the structure aspect of programming language. We then enhance the representations via contrastive learning with hard negative and hard positive constructed in an unsupervised manner.
  arXiv  Detail & Related papers  (2024-02-02T22:19:15Z)
• In-Context Learning for MIMO Equalization Using Transformer-Based Sequence Models [44.161789477821536]
  Large pre-trained sequence models have the capacity to carry out in-context learning (ICL) In ICL, a decision on a new input is made via a direct mapping of the input and of a few examples from the given task. We demonstrate via numerical results that transformer-based ICL has a threshold behavior.
  arXiv  Detail & Related papers  (2023-11-10T15:09:04Z)
• Towards Understanding How Transformers Learn In-context Through a Representation Learning Lens [9.590540796223715]
  In this paper, we attempt to explore the in-context learning process in Transformers through a lens of representation learning. The ICL inference process of the attention layer aligns with the training procedure of its dual model, generating token representation predictions. We extend our theoretical conclusions to more complicated scenarios, including one Transformer layer and multiple attention layers.
  arXiv  Detail & Related papers  (2023-10-20T01:55:34Z)
• How Do Transformers Learn In-Context Beyond Simple Functions? A Case Study on Learning with Representations [98.7450564309923]
  This paper takes initial steps on understanding in-context learning (ICL) in more complex scenarios, by studying learning with representations. We construct synthetic in-context learning problems with a compositional structure, where the label depends on the input through a possibly complex but fixed representation function. We show theoretically the existence of transformers that approximately implement such algorithms with mild depth and size.
  arXiv  Detail & Related papers  (2023-10-16T17:40:49Z)
• A Meta-Learning Perspective on Transformers for Causal Language Modeling [17.293733942245154]
  The Transformer architecture has become prominent in developing large causal language models. We establish a meta-learning view of the Transformer architecture when trained for the causal language modeling task. Within the inner optimization, we discover and theoretically analyze a special characteristic of the norms of learned token representations within Transformer-based causal language models.
  arXiv  Detail & Related papers  (2023-10-09T17:27:36Z)
• 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)
• What and How does In-Context Learning Learn? Bayesian Model Averaging, Parameterization, and Generalization [111.55277952086155]
  We study In-Context Learning (ICL) by addressing several open questions. We show that, without updating the neural network parameters, ICL implicitly implements the Bayesian model averaging algorithm. We prove that the error of pretrained model is bounded by a sum of an approximation error and a generalization error.
  arXiv  Detail & Related papers  (2023-05-30T21:23:47Z)
• Unveiling Transformers with LEGO: a synthetic reasoning task [23.535488809197787]
  We study how the transformer architecture learns to follow a chain of reasoning. In some data regime the trained transformer finds "shortcut" solutions to follow the chain of reasoning. We find that one can prevent such shortcut with appropriate architecture modification or careful data preparation.
  arXiv  Detail & Related papers  (2022-06-09T06:30:17Z)
• Thinking Like Transformers [64.96770952820691]
  We propose a computational model for the transformer-encoder in the form of a programming language. We show how RASP can be used to program solutions to tasks that could conceivably be learned by a Transformer. We provide RASP programs for histograms, sorting, and Dyck-languages.
  arXiv  Detail & Related papers  (2021-06-13T13:04:46Z)

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.