INDEX

Explanations

actions involving physical interactions or accidents

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

accidental physical events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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

isOpen

-0.35

GO

-0.31

 location

-0.31

sur

-0.31

sur

-0.31

 наси

-0.30

望

-0.30

filepath

-0.30

parentNode

-0.30

GO

-0.29

POSITIVE LOGITS

URLException

0.68

ScopeManager

0.66

sizeCache

0.63

 wireType

0.60

 onOptions

0.59

 accidentally

0.56

 AttributeSet

0.56

propTypes

0.56

QMetaType

0.54

InitVars

0.53

Activations Density 0.018%

actions involving physical interactions or accidents

accidental physical events

No Comments

No Known Activations