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

raine

-0.75

é¾

-0.73

ulia

-0.72

plant

-0.70

ãĤ¼ãĤ¦ãĤ¹

-0.69

DNA

-0.68

adelphia

-0.68

Song

-0.67

gypt

-0.66

uron

-0.66

POSITIVE LOGITS

 pains

0.73

 Jacobs

0.71

 Cullen

0.68

 queues

0.67

 Stra

0.66

 Purg

0.64

eries

0.63

 McMaster

0.62

thodox

0.60

loads

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.