INDEX

Explanations

words related to entities or aspects connected to the prefix "Er."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 17-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_17/width_65k/average_l0_125

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.17.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

 الحره

-0.88

 nakalista

-0.85

httphttps

-0.84

tvguidetime

-0.82

 noDo

-0.79

ViewFeatures

-0.77

AxisAlignment

-0.73

ollectionView

-0.73

WillAppear

-0.72

 "..\..\..\

-0.71

POSITIVE LOGITS

Er

0.69

Er

0.67

eru

0.64

er

0.63

Eri

0.62

Ers

0.59

 Ernie

0.59

 Enfin

0.58

 Eras

0.57

Ero

0.56

Activations Density 0.244%

words related to entities or aspects connected to the prefix "Er."

No Comments

No Known Activations