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

 Railway

-0.63

 Daylight

-0.59

 Diet

-0.58

 Invaders

-0.57

 Bastard

-0.56

 Clown

-0.56

elope

-0.56

 Brewer

-0.56

 Rudolph

-0.55

Chief

-0.55

POSITIVE LOGITS

ogy

0.84

arer

0.81

usk

0.75

Ñı

0.75

ograp

0.73

Americ

0.72

IAS

0.71

,,,,

0.70

ynes

0.70

@#&

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.