INDEX

Explanations

pairs

np_max-act · gemini-2.0-flash

This neuron activates on instructions or contexts about “question and answer pairs,” i.e., it detects mentions of generating Q&A pair lists.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Alarm

-0.08

암

-0.07

SECRET

-0.07

Υ

-0.07

盟

-0.07

РН

-0.07

форт

-0.06

IOC

-0.06

prak

-0.06

ención

-0.06

POSITIVE LOGITS

 coupling

0.08

 cries

0.07

 backdrop

0.06

新

0.06

 Observatory

0.06

cratch

0.06

 Κο

0.06

 coach

0.06

/cart

0.06

 coupled

0.06

Activations Density 0.008%

pairs

This neuron activates on instructions or contexts about “question and answer pairs,” i.e., it detects mentions of generating Q&A pair lists.

No Comments

No Known Activations