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

Tau

-0.77

ividual

-0.75

elope

-0.75

ailability

-0.75

uctor

-0.75

razil

-0.74

arez

-0.73

ership

-0.73

iameter

-0.70

ickson

-0.70

POSITIVE LOGITS

 Remastered

0.70

²¾

0.68

 disapp

0.67

 homework

0.65

 recre

0.65

?).

0.65

asures

0.65

kered

0.65

Ŀ

0.64

Ì¶

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.