INDEX

Explanations

terms related to grace and headings in text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_0/width_65k/average_l0_73

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 resourceCulture

-0.63

 lipat

-0.61

 Preferencias

-0.60

AndEndTag

-0.59

rões

-0.57

 sauvage

-0.56

WithMany

-0.56

DotNetBar

-0.56

 religieuses

-0.56

 Bradbury

-0.56

POSITIVE LOGITS

 Heading

0.71

Marshal

0.66

 caption

0.66

 download

0.65

TEntity

0.63

 tumb

0.63

 Download

0.62

Kt

0.60

 marshal

0.58

esthes

0.58

Activations Density 0.045%

terms related to grace and headings in text

No Comments

No Known Activations