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

andum

-0.74

han

-0.71

 Turks

-0.70

bag

-0.70

onite

-0.69

itbart

-0.69

tics

-0.65

on

-0.65

afa

-0.64

eh

-0.63

POSITIVE LOGITS

=~

0.80

eatures

0.78

sburgh

0.74

cedes

0.68

 reefs

0.66

enses

0.65

pse

0.65

 camoufl

0.64

 Lakes

0.63

 arrang

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.