INDEX

Explanations

phrases that indicate actions or instructions related to viewing or enlarging content

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

commands related to user interactions, specifically those involving expanding or enlarging content

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

to [action verb]

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

 AppDelegate

-0.65

 hålla

-0.65

المناصب

-0.64

ViewFeatures

-0.63

 mergeFrom

-0.61

FunctionFlags

-0.61

rungsseite

-0.61

bpy

-0.60

LLocation

-0.60

XMLSchema

-0.59

POSITIVE LOGITS

 view

0.75

 access

0.68



0.68

see

0.65

 visit

0.63

usercontent

0.61

 read

0.58

ViewImports

0.58

 Access

0.56

 acess

0.53

Activations Density 0.018%

phrases that indicate actions or instructions related to viewing or enlarging content

commands related to user interactions, specifically those involving expanding or enlarging content

to [action verb]

No Comments

No Known Activations