INDEX

Explanations

phrases and words that indicate emphasis or significance through bold formatting

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

bold text or emphasis

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

principalColumn

-0.59

userManager

-0.55

帖最后由

-0.53

 VIDEOTAPE

-0.52

 ویکی‌پدیای

-0.52

 mxArray

-0.51

SessionFactory

-0.51

حياته

-0.50

UAR

-0.50

 Infórmanos

-0.49

POSITIVE LOGITS

 Bold

1.12

Bold

1.08

 bold

1.01

 BOLD

0.97

bold

0.94

BOLD

0.82

 boldness

0.71

 bolder

0.70

 boldly

0.67

setBold

0.67

Activations Density 0.004%

phrases and words that indicate emphasis or significance through bold formatting

bold text or emphasis

No Comments

No Known Activations