INDEX

Explanations

text formatting tags and indicators related to document structure

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

if clauses

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



-0.86

uxxxx

-0.65

oredCriteria

-0.63

Autoritní

-0.62

esModule

-0.60



-0.58

$_['

-0.57

 arşivlendi

-0.56

 تعدى

-0.54

RegressionTest

-0.54

POSITIVE LOGITS

Content

0.45

<i>

0.40

content

0.39

CONTENT

0.38

XmlElement

0.35

Véxase

0.35

<em>

0.35

 contenu

0.35

 content

0.34

 écrits

0.33

Activations Density 0.019%

text formatting tags and indicators related to document structure

if clauses

No Comments

No Known Activations