INDEX

Explanations

numerical or coded representations of data

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eseorang

-0.53

jahtera

-0.40

 pezzi

-0.37

 Colour

-0.36

 thereon

-0.36

 retten

-0.35

 magazine

-0.34

melidir

-0.34

 esterna

-0.34

 termica

-0.34

POSITIVE LOGITS

Portale

0.88

Portail

0.84

Personendaten

0.78

Portály

0.70

PhysRev

0.68

----</

0.65

بوابة

0.61

AnchorStyles

0.59

 beginnetje

0.59

 comuniques

0.59

Activations Density 0.001%

numerical or coded representations of data

No Comments

No Known Activations