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

 tangled

-0.73

 Warfare

-0.73

iaries

-0.69

 entangled

-0.67

 Thread

-0.67

 Universe

-0.64

 Shield

-0.61

mal

-0.61

 Boxing

-0.61

 spurious

-0.60

POSITIVE LOGITS

 reluct

0.78

ftime

0.74

 rall

0.74

bably

0.73

pez

0.70

rett

0.70

atra

0.69

GoldMagikarp

0.69

 accordingly

0.65

yip

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.