INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Hancock

-0.69

ilation

-0.69

 hardened

-0.67

 Prairie

-0.67

inguishable

-0.66

 Superman

-0.63

 Randolph

-0.63

ooth

-0.62

 Marcos

-0.60

KGB

-0.60

POSITIVE LOGITS

pent

0.70

pa

0.70

Mods

0.70

ãĥĨ

0.69

voc

0.68

VW

0.65

 mysql

0.65

bh

0.65

lehem

0.64

nu

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.