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

=]

-0.93

asio

-0.81

onday

-0.77

Catal

-0.76

phabet

-0.76

isSpecialOrderable

-0.74

 Grateful

-0.73

agos

-0.72

soDeliveryDate

-0.71

orsche

-0.68

POSITIVE LOGITS

phan

0.72

GH

0.72

 stocking

0.65

vag

0.64

vag

0.63

cker

0.62

 quarantine

0.62

 multiplier

0.61

 Hyper

0.61

gery

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.