INDEX

Explanations

linguistic patterns and structures related to actions and processes

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

телен

-0.31

står

-0.30

 Oliven

-0.28

站

-0.27

 λογ

-0.27

 Physiological

-0.26

be

-0.26

패

-0.25

Popis

-0.25

 qued

-0.25

POSITIVE LOGITS

awtextra

0.75

 للاسماء

0.73

 queſta

0.62

 initComponents

0.61

المشاركات

0.61

featureID

0.60

 ourselves

0.60

 vectorielle

0.60

 kasarigan

0.59

Vidite

0.59

Activations Density 0.204%

linguistic patterns and structures related to actions and processes

No Comments

No Known Activations