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

byss

-0.75

asses

-0.73

hid

-0.73

olen

-0.70

 concealed

-0.67

perse

-0.65

resses

-0.65

cb

-0.63

umbledore

-0.63

assed

-0.62

POSITIVE LOGITS

lance

0.78

ismo

0.72

 Lennon

0.67

 suprem

0.67

atile

0.66

 cardinal

0.65

etary

0.64

 Boxing

0.64

 Cathedral

0.62

Vaj

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.