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

igate

-0.72

quit

-0.66

omination

-0.65

iquette

-0.65

lance

-0.64

jriwal

-0.62

 Pend

-0.61

ential

-0.61

 Micha

-0.61

anasia

-0.61

POSITIVE LOGITS

 Bild

0.73

hog

0.72

las

0.70

 Skydragon

0.67

Football

0.67

PHOTOS

0.65

UGH

0.65

hens

0.62

raq

0.61

Cra

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.