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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wherein

-0.86

utra

-0.84

 Skydragon

-0.82

ensis

-0.79

opsis

-0.79

 whereby

-0.74

stice

-0.73

 Protestants

-0.67

uous

-0.65

ypes

-0.64

POSITIVE LOGITS

 Belt

0.82

lynn

0.73

 arms

0.72

 workforce

0.70

ses

0.70

 electorate

0.69

 neighbour

0.68

 citiz

0.68

 orderly

0.67

own

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.