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

 dining

-0.62

 sourcing

-0.60

 facult

-0.60

idas

-0.59

 prud

-0.58

 tacit

-0.58

 manag

-0.58

 regulatory

-0.57

 istg

-0.57

 obliged

-0.57

POSITIVE LOGITS

llah

0.76

osure

0.72

yout

0.70

OOL

0.70

EVA

0.70

alid

0.70

rosso

0.69

aez

0.69

brance

0.67

igil

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.