INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Seym

-0.72

hammad

-0.66

iets

-0.64

 Angelo

-0.64

 Hodg

-0.63

 infancy

-0.63

 Canaver

-0.63

ypes

-0.63

 Call

-0.60

 holidays

-0.60

POSITIVE LOGITS

vous

0.77

Unix

0.76

Ã©t

0.75

Tokens

0.74

NP

0.69

WI

0.69

MQ

0.69

forth

0.69

NAT

0.68

à¨

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.