INDEX

Explanations

warning signals or indications of caution

oai_token-act-pair · gpt-4o-mini Triggered by @bot

warning signs or alerts

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

">$

-0.45

tub

-0.43

plu

-0.42

stateProvider

-0.41

ITURE

-0.40

lorette

-0.39

Simply

-0.39

Plu

-0.38

Cole

-0.38

Plu

-0.37

POSITIVE LOGITS

 warning

1.23

 Warning

1.17

 warnings

1.15

warning

1.09

Warnings

1.04

 Warnings

1.04

warnings

1.00

Warning

0.99

warn

0.99

 warn

0.98

Activations Density 0.009%

warning signals or indications of caution

warning signs or alerts

No Comments

No Known Activations