INDEX

Explanations

guessing games

np_max-act · gemini-2.0-flash

This neuron fires on the formatted heading and instruction tokens used in contest or game prompts (e.g. “Guess,” “the,” numbered titles, and other call-to-action headings).

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

Negative Logits

۱۹۴

-0.07

noticed

-0.06

pter

-0.06

уже

-0.06

/wiki

-0.06

orld

-0.06

 Sense

-0.06

 tức

-0.06

те

-0.06

lx

-0.06

POSITIVE LOGITS

 progressDialog

0.07

.ADMIN

0.07

 imageURL

0.07

=?",

0.06

丈夫

0.06

.Seek

0.06

 représent

0.06

(opcode

0.06

/Branch

0.06

)=(

0.06

Activations Density 0.120%

guessing games

This neuron fires on the formatted heading and instruction tokens used in contest or game prompts (e.g. “Guess,” “the,” numbered titles, and other call-to-action headings).

No Known Activations

guessing games

This neuron fires on the formatted heading and instruction tokens used in contest or game prompts (e.g. “Guess,” “the,” numbered titles, and other call-to-action headings).

No Known Activations