INDEX

Explanations

terms indicating prior names or previous identities of entities

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

past states or names

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

:+:

-0.47

 poil

-0.43

Abp

-0.42

 ✭✭

-0.41

 exposure

-0.38

ɵɵ

-0.38

 opportunity

-0.36

 issue

-0.36

nestjs

-0.36

 exercise

-0.36

POSITIVE LOGITS

CloseOperation

0.67

帖最后由

0.66

AnchorStyles

0.56

fjspx

0.55

ंदीखरीदारी

0.54

setViewportView

0.53

formerly

0.52

 renaming

0.50

sendRedirect

0.50

 pungkas

0.49

Activations Density 0.010%

terms indicating prior names or previous identities of entities

past states or names

No Comments

No Known Activations