INDEX

Explanations

words indicating quantity or variation

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

in some cases

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

 nonUne

-0.70

 increí

-0.70

 Theſe

-0.68

 himſelf

-0.68

 estadounid

-0.63

 müſſen

-0.62

 Verſ

-0.62

 desmotivaciones

-0.60

antaranya

-0.60

 Beſ

-0.60

POSITIVE LOGITS

in

0.83

In

0.73

IN

0.58

In

0.54

 وفي

0.51

 accordance

0.50

 spite

0.47

 isIn

0.47

 vicinity

0.45

 manner

0.45

Activations Density 0.069%

words indicating quantity or variation

in some cases

No Comments

No Known Activations