Learn Before
Evaluating Attention Mechanism Stability
An AI research team is comparing two different model architectures for summarizing very long legal documents. They observe the following behaviors in the attention mechanism:
- Model Alpha: When processing a 20,000-token document, its attention mechanism frequently produces highly concentrated weight distributions. For many output tokens, a single input token receives an attention score greater than 0.9, while all other tokens receive scores close to zero.
- Model Beta: When processing the same document, its attention mechanism produces more distributed weights. While specific input tokens are still highlighted, a consistent portion of the attention weight (e.g., 15-20%) is always allocated to a small, fixed set of tokens that are accessible from every position in the sequence.
Based on these observations, which model is more likely to have a stable training process, and why? Justify your answer by explaining the impact of the observed attention distribution on model learning.
0
1
Tags
Ch.2 Generative Models - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Evaluation in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
An engineer is training a model on extremely long text sequences and notices that the training process is unstable. Upon inspection, they find that the attention mechanism frequently assigns nearly all of its weight to a single, specific token, ignoring the rest of the context. This extreme concentration of attention scores is causing gradients to become erratic. Which of the following strategies directly addresses this problem by encouraging a less concentrated, more stable distribution of atte
A machine learning engineer is training a model on very long documents and observes that the attention mechanism is behaving unstably: for some parts of the text, the attention weights are extremely concentrated on a single word, while for others, they are so spread out they are effectively zero. After implementing a change, the engineer observes that the attention weights are now more smoothly distributed, leading to more consistent model performance. Which of the following changes best explain
Evaluating Attention Mechanism Stability