INDEX

Explanations

enjoyed

np_max-act · gemini-2.0-flash

The neuron strongly responds to the assistant’s closing or sign-off phrases (e.g. “I hope you enjoyed this episode…”).

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

 deze

-0.06

 день

-0.06

ّة

-0.06

 Lucia

-0.06

ک

-0.06

odic

-0.06

(async

-0.06

Ố

-0.06

者

-0.06

/icons

-0.06

POSITIVE LOGITS

([$

0.08

DataReader

0.07

 Meal

0.07

 getElement

0.06

firefox

0.06

Michelle

0.06

retrieve

0.06

allowed

0.06

 serif

0.06

('_

0.06

Activations Density 0.029%

enjoyed

The neuron strongly responds to the assistant’s closing or sign-off phrases (e.g. “I hope you enjoyed this episode…”).

No Comments

No Known Activations