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

itual

-0.67

oire

-0.66

 lash

-0.64

lf

-0.63

elight

-0.62

iday

-0.61

naires

-0.60

usra

-0.60

 itch

-0.60

ĸļ

-0.59

POSITIVE LOGITS

aucus

0.87

ucker

0.70

 Complex

0.70

*/(

0.69

/"

0.68

angan

0.68

plex

0.67

anski

0.66

 twent

0.64

Exc

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.