INDEX

Explanations

occurrences of mentioning and referencing prior content or context

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

referencing previously presented information

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

 rates

-0.29

特

-0.26

ierta

-0.26

 statt

-0.25

 than

-0.25

<0xDA>

-0.25

Inside

-0.24

 Reichen

-0.24

now

-0.24

ITAT

-0.23

POSITIVE LOGITS

 Италијани

0.77

Derbyniad

0.74

 snippetHide

0.74

 Infórmanos

0.73

 مشين

0.73

DockStyle

0.69

 виправивши

0.69

UnknownFieldSet

0.68

contentLoaded

0.68

 Normdatei

0.68

Activations Density 0.066%

occurrences of mentioning and referencing prior content or context

referencing previously presented information

No Comments

No Known Activations