INDEX

Explanations

references to specific names, titles, or key identifiers in various contexts

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

code placeholders and 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

GEBURTSDATUM

-0.40

ույ

-0.39

nthetic

-0.39

ViewFeatures

-0.39

elemField

-0.39

urement

-0.38

 Walkover

-0.37

nextProps

-0.37

klo

-0.37

دانشنامهٔ

-0.37

POSITIVE LOGITS

期刊论文

0.58

 nahilalakip

0.55

quelize

0.49

########.

0.49

eseorang

0.45

 Audiodateien

0.45

 تعدى

0.40

()}>

0.39

 faſt

0.38

}.

0.38

Activations Density 0.071%

references to specific names, titles, or key identifiers in various contexts

code placeholders and names

No Comments

No Known Activations