INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

avorite

-0.78

 "$:/

-0.76

 VIDEOS

-0.76

VERTISEMENT

-0.74

udos

-0.72

BLIC

-0.70

iculty

-0.70

rencies

-0.68

erella

-0.67

 corrid

-0.66

POSITIVE LOGITS

 dispers

0.67

 squad

0.64

 defunct

0.61

roll

0.60

Kal

0.59

bows

0.58

ãĤ¤ãĥĪ

0.57

 Explosive

0.57

orthern

0.57

amination

0.56

Activations Density 0.000%

No Known Activations

This feature has no known activations.