Finding patterns in Knowledge Attribution for Transformers
share
We analyze the Knowledge Neurons framework for the attribution of factual and relational knowledge to particular neurons in the transformer network. We use a 12-layer multi-lingual BERT model for our experiments. Our study reveals various interesting phenomena. We observe that mostly factual knowledge can be attributed to middle and higher layers of the network(≥ 6). Further analysis reveals that the middle layers(6-9) are mostly responsible for relational information, which is further refined into actual factual knowledge or the "correct answer" in the last few layers(10-12). Our experiments also show that the model handles prompts in different languages, but representing the same fact, similarly, providing further evidence for effectiveness of multi-lingual pre-training. Applying the attribution scheme for grammatical knowledge, we find that grammatical knowledge is far more dispersed among the neurons than factual knowledge.
READ FULL TEXT
