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

ivating

-0.78

catentry

-0.75

aci

-0.74

odynam

-0.74

onom

-0.73

unning

-0.72

ï¿½ï¿½ï¿½ï¿½

-0.71

cru

-0.71

ãĥĥãĥĪ

-0.69

itsch

-0.68

POSITIVE LOGITS

 Salvation

0.64

 Pentagon

0.63

 mail

0.60

 detachment

0.59

cub

0.58

ya

0.58

 paste

0.58

 footnote

0.57

gar

0.57

 gelatin

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.