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

AAF

-0.67

gap

-0.66

affe

-0.65

KB

-0.63

emphasis

-0.62

spin

-0.60

 Dictionary

-0.60

bie

-0.60

eele

-0.60

Mp

-0.59

POSITIVE LOGITS

wcsstore

0.98

imes

0.69

hon

0.68

 Marketable

0.67

ulkan

0.66

Initialized

0.66

 redeemed

0.66

ertodd

0.65

coh

0.64

İ

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.