INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rency

-0.76

 guiActiveUnfocused

-0.75

holes

-0.69

ãĥĨãĤ£

-0.69

riott

-0.68

 Flavoring

-0.66

udic

-0.65

rils

-0.65

hole

-0.65

Events

-0.65

POSITIVE LOGITS

cknow

0.72

Yon

0.67

 Republic

0.65

 guid

0.64

ĻĤ

0.60

vernment

0.60

\">

0.59

alach

0.58

Rep

0.57

fug

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.