INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥĲ

-0.76

eur

-0.74

awa

-0.72

èĢ

-0.67

oooooooooooooooo

-0.64

////////////////

-0.64

BILITIES

-0.64

ocard

-0.63

à¤

-0.63

leness

-0.63

POSITIVE LOGITS

iannopoulos

0.87

 Pradesh

0.69

adesh

0.67

 Shade

0.66

anooga

0.66

agos

0.64

hire

0.64

hak

0.63

iann

0.63

»Ĵ

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.