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

NetMessage

-0.74

IX

-0.73

lance

-0.70

irlf

-0.68

Dispatch

-0.68

ioxide

-0.65

eye

-0.65

Eye

-0.65

invest

-0.64

arb

-0.64

POSITIVE LOGITS

ascript

0.78

atories

0.73

 supra

0.69

 masc

0.68

querque

0.66

 Abbott

0.65

{"

0.65

Pax

0.63

zin

0.62

cffff

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.