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

asury

-0.71

med

-0.68

ablishment

-0.67

rain

-0.67

icum

-0.66

inyl

-0.66

uliffe

-0.64

andals

-0.64

osher

-0.64

itated

-0.63

POSITIVE LOGITS

tnc

0.72

Tee

0.65

CVE

0.65

 Jacqu

0.65

 perjury

0.64

quickShipAvailable

0.64

tee

0.64

Kev

0.63

 Vaughn

0.63

 disbel

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.