INDEX

Explanations

names of people and entities associated with family or relationships

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

EndInit

-0.65

quelize

-0.55

 retired

-0.54

EndGlobalSection

-0.53

SequentialGroup

-0.50

退休

-0.49

WindowConstants

-0.48

rhosis

-0.48

 ancianos

-0.47

 Retired

-0.47

POSITIVE LOGITS

 interns

0.57

 Danielle

0.57

Jessica

0.54

 Jessica

0.54

Danielle

0.53

 Micah

0.53

Travis

0.52

 Alyssa

0.52

 Erin

0.52

 Josh

0.52

Activations Density 0.304%

names of people and entities associated with family or relationships

No Comments

No Known Activations