INDEX

Explanations

first-person expressions of confusion or uncertainty

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

 MessageBoxIcon

-0.43

 وكانت

-0.41

väl

-0.41

usda

-0.40

 connais

-0.40

Gi

-0.40

 spectrum

-0.39

შ

-0.38

matchers

-0.38

protective

-0.38

POSITIVE LOGITS

 دیکھیے

0.86

зулта

0.73

AsUp

0.71

 EconPapers

0.69

Koordinaten

0.69

WillAppear

0.68

AndEndTag

0.68

 kaarangay

0.68

 Italijanski

0.66

MessageOf

0.66

Activations Density 0.034%

first-person expressions of confusion or uncertainty

No Comments

No Known Activations