INDEX

Explanations

common terms and identifiers related to various scientific and technical contexts

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

hexadecimal number

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

 knock

-0.40

knock

-0.38

dade

-0.36

 dieß

-0.36

 whereupon

-0.36

noticed

-0.35

sne

-0.35

각주

-0.35

dL

-0.35

sof

-0.35

POSITIVE LOGITS

 AssemblyCulture

0.56

 informée

0.53

Personensuche

0.52

Tembelea

0.52

yarnpkg

0.52

 '\\;'

0.51

TargetException

0.49

__':

0.47

 مرئيه

0.47

 חיצוניים

0.47

Activations Density 0.078%

common terms and identifiers related to various scientific and technical contexts

hexadecimal number

No Comments

No Known Activations