INDEX

Explanations

references to individuals and groups within various contexts

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

 gobier

-0.41

 souverain

-0.40

 Komunikasi

-0.40

 døde

-0.40

 tambi

-0.40

 ayude

-0.38

 stöd

-0.38

 trámite

-0.37

 Keuangan

-0.37

 fijo

-0.37

POSITIVE LOGITS

 being

0.71

rungsseite

0.62

 surla

0.61

uxxxx

0.57

Kjelder

0.56

being

0.54

OGND

0.53

 breaking

0.52

digheden

0.52

 Meksiku

0.51

Activations Density 0.057%

references to individuals and groups within various contexts

No Comments

No Known Activations