INDEX

Explanations

proper nouns, particularly locations and names

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

city 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

 cherchés

-0.57

Rüyada

-0.57

 esternos

-0.56

 otomatig

-0.51

Ӕ

-0.50

Hochspringen

-0.49

Notae

-0.48

 autorytatywna

-0.47



-0.47

oa̍t

-0.47

POSITIVE LOGITS

DeleteBehavior

0.46

 radi

0.36

imp

0.36

Ars

0.36

 Mercu

0.35

reprodu

0.34

 Beverly

0.34

 clad

0.33

デア

0.33

drawSprites

0.33

Activations Density 0.009%

proper nouns, particularly locations and names

city names

No Comments

No Known Activations