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

ulty

-0.82

orescence

-0.75

ipolar

-0.74

icago

-0.70

erva

-0.70

perial

-0.68

terness

-0.67

 horizont

-0.67

ults

-0.66

auga

-0.66

POSITIVE LOGITS

CHAT

0.70

SourceFile

0.70

 Explain

0.69

Hell

0.66

atl

0.63

Question

0.63

Has

0.63

Null

0.63

Send

0.62

Tax

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.