INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

sbm

-0.88

ãĤ³

-0.84

cair

-0.82

avery

-0.74

\">

-0.74

schild

-0.73

oldown

-0.67

hur

-0.67

ãĥ³

-0.66

achus

-0.66

POSITIVE LOGITS

Interstitial

0.68

Duo

0.67

 Twist

0.65

 Waterloo

0.65

 Alph

0.64

RIP

0.63

 Doll

0.62

 Warrant

0.61

 Geek

0.60

 Unic

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.