INDEX

Explanations

references to legal charges or legal terminology

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

-0.68

 jScrollPane

-0.65

Diweddarwch

-0.65

aryen

-0.63

 مرئيه

-0.62

__((

-0.62

 saites

-0.61

 Reverso

-0.61

ExecuteReader

-0.60

expandindo

-0.60

POSITIVE LOGITS

SequentialGroup

0.57

 colectiva

0.55

 Coulomb

0.53

 répon

0.52

 "..\..\

0.50

 eléctricos

0.47

ateľ

0.46

 Dica

0.46

 pezzi

0.45

 τῶν

0.45

Activations Density 0.129%

references to legal charges or legal terminology

No Comments

No Known Activations