INDEX

Explanations

references to statistical data and results from research studies

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

 cref

-0.60

ObjectModel

-0.54

 siste

-0.53

 disponibilités

-0.51

mberg

-0.51

+#+

-0.49

setRotation

-0.49

atech

-0.49

-0.48

setPath

-0.48

POSITIVE LOGITS

 itſelf

0.89

 themſelves

0.73

脚注の使い方

0.72

 greateſt

0.71

 Roskov

0.69

ſelves

0.67

 houſe

0.66

 muſt

0.66

 ſhould

0.66

 myſelf

0.65

Activations Density 0.120%

references to statistical data and results from research studies

No Comments

No Known Activations