INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

uffed

-0.76

raft

-0.74

nen

-0.72

NN

-0.71

abolic

-0.68

 overpowered

-0.68

aiden

-0.67

iencies

-0.67

hew

-0.66

ovi

-0.65

POSITIVE LOGITS

 Aven

0.69

 satir

0.64

debian

0.64

 diseng

0.64

resp

0.64

vity

0.62

 partName

0.62

 externalToEVAOnly

0.61

 Corbyn

0.61

 volcan

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.