INDEX

Explanations

phrases indicating relationships between quantity, quality, and user input

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

more longer

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

 للمعارف

-0.49

VideoCapture

-0.44

ScopeManager

-0.44

rapeau

-0.42

<()>

-0.42

mister

-0.41

initWithFrame

-0.41

 HttpNotFound

-0.41

PHeader

-0.41

ruptedException

-0.41

POSITIVE LOGITS

 كومونز

0.45

 càng

0.45

越多

0.44

越是

0.40

 esboço

0.39

oa̍t

0.36

 onData

0.36

 duradero

0.35

越好

0.35

 naciones

0.35

Activations Density 0.028%

phrases indicating relationships between quantity, quality, and user input

more longer

No Comments

No Known Activations