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

glers

-0.71

ming

-0.67

mes

-0.66

ropy

-0.65

tid

-0.63

ertodd

-0.63

Forest

-0.63

 entropy

-0.63

cling

-0.63

cdn

-0.62

POSITIVE LOGITS

 misunder

0.83

everal

0.71

illet

0.70

emale

0.68

EMBER

0.68

GUN

0.67

REE

0.65

{\

0.63

avorite

0.63

ethe

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.