INDEX

Explanations

phrases indicating significance or importance of concepts or items

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

one of the most

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

 pinulongan

-0.51

 carc

-0.48

GeneratedCode

-0.47

dac

-0.47

parseFrom

-0.47

 BaseActivity

-0.47

 insp

-0.47

 xvii

-0.46

gor

-0.45

ollectionView

-0.45

POSITIVE LOGITS

 satunya

0.56

 antaranya

0.52

----</

0.50

 tärke

0.46

 متعلقه

0.45

 nôtre

0.45

 casada

0.45

 solches

0.43

 esimer

0.43

 yakni

0.43

Activations Density 0.021%

phrases indicating significance or importance of concepts or items

one of the most

No Comments

No Known Activations