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

lda

-0.69

utral

-0.67

ewitness

-0.65

xes

-0.65

instein

-0.64

 bush

-0.62

selves

-0.62

leck

-0.61

avior

-0.61

 degradation

-0.60

POSITIVE LOGITS

actionDate

0.97

mable

0.71

 sadd

0.66

 Pathfinder

0.64

Drive

0.64

atar

0.63

iator

0.61

lehem

0.60

Offline

0.59

albeit

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.