INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Wars

-0.92

mble

-0.69

éŃĶ

-0.69

SpaceEngineers

-0.67

 Cruise

-0.64

ansk

-0.62

 Britann

-0.62

WAR

-0.62

Mad

-0.62

 Wales

-0.61

POSITIVE LOGITS

 suspic

0.88

friend

0.82

friends

0.73

pee

0.70

zona

0.69

erville

0.67

imentary

0.66

 crypt

0.65

handler

0.65

olicited

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.