Problem
Research questions and friction points this paper is trying to address.
attention heads
Context-aware Machine Translation
causal analysis
general-purpose
Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation
🤖 AI Summary
本文通过梯度归因策略分析注意力头，解决上下文感知机器翻译中的歧义问题，揭示了多功能注意力头对模型性能的影响。
📝 Abstract
In this paper, we introduce a gradient-based head attribution strategy where the Token-level Max-Margin loss is backpropagated to the attention maps. This framework enables a large-scale causal analysis of attention heads, making it suitable for LLMs. We evaluate our method on the task of disambiguation in Context-aware Machine Translation, where we analyze 50 phenomena across 4 models and 4 language directions. We empirically show the alignment of our method with the effects of increasing the attention scores of token-to-token relations on three models and two language directions, ensuring the robustness of our method. Our analysis reveals the presence of the "general-purpose" attention heads that improve the model's performance when attending to different relations. We find that the average attention a head assigns to a relation does not necessarily relate to the model's performance, which suggests that the models developed redundancies during training in terms of the head functions.
Innovation
Methods, ideas, or system contributions that make the work stand out.
gradient-based head attribution
Token-level Max-Margin loss
large-scale causal analysis
general-purpose attention heads
🔎 Similar Papers
No similar papers found.
