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

iver

-0.79

ãĥīãĥ©

-0.79

rosso

-0.77

irie

-0.77

amacare

-0.76

orman

-0.75

atches

-0.75

ranch

-0.74

rador

-0.73

izon

-0.73

POSITIVE LOGITS

mun

0.65

GP

0.64

 disg

0.63

 includ

0.63

âĤ¬

0.63

 generously

0.62

 Judicial

0.62

GRE

0.60

yea

0.59

 Dice

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.