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

":["

-0.85

 Subway

-0.72

(%)

-0.69

existent

-0.67

AIR

-0.66

hab

-0.65

 Rooney

-0.64

STER

-0.63

 Sikh

-0.63

ãĥ¥

-0.61

POSITIVE LOGITS

©¶æ

0.96

hiba

0.73

Arc

0.71

à¼

0.71

 chapter

0.69

 inference

0.69

Arc

0.68

ħĭ

0.68

amina

0.67

umn

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.