INDEX

Explanations

HTML-like tags and formatting elements in the text

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

technical delimiters and code symbols

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

 saurait

-0.46

 Administrativna

-0.43

ดาว

-0.42

 asunder

-0.42

chaub

-0.42

 Mühe

-0.40

zni

-0.40

böz

-0.40

 Trio

-0.40

POSITIVE LOGITS

rungsseite

0.76

 resourceCulture

0.63

ValueStyle

0.59

 disambiguazione

0.57

MessageState

0.55

enterOuterAlt

0.53

IsMutable

0.50

 Roskov

0.47

 CreateTagHelper

0.46

httphttps

0.46

Activations Density 0.001%

HTML-like tags and formatting elements in the text

technical delimiters and code symbols

No Comments

No Known Activations