IC-722Entity tracking in Llama-7b is implemented by detecting and transmitting the positional information of the correct entity, with distinct head groups for position detection, transmission, and value fetching

Nikhil Prakash, Tamar Rott Shaham, Tal Haklay, Yonatan Belinkov, David Bau

SourceFine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking

Using desiderata-based component masking (DCM), the authors show that 71.43% of group B heads and 70.0% of group C heads encode the positional information of the correct object, while all group A heads encode the object value. Activation patching confirms that group A (value fetcher) retrieves the object at the position identified by groups B and C (position detector and transmitter). This position-based scheme is the same in Vicuna-7b and Goat-7b, confirming that fine-tuning does not change the mechanism, only enhances it.

Evidence
interventional
Key metric
71.43% of group B heads and 70.0% of group C heads encode positional information (DCM); group A heads primarily encode object value; position detector and transmitter accuracy similar across Llama-7b, Vicuna-7b, Goat-7b
Caveat
The exact nature of the positional information is not fully characterized; the authors note that additional segments at the start and boxes before the correct segment interfere with the positional encoding, suggesting relative position from the start token.
Model
LLaMA Llama 7B, Vicuna Vicuna-7B, GOAT-7B
Methods
Activation patching / Activation replacement / Cross-model activation patching (CMap) [primary]
Related work
Davies et al. 2023 (DCM) [builds-on]
Related findings
IC-721, IC-723
Extraction
automatic-extraction