INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

mit

-0.85

pi

-0.73

hered

-0.67

 chau

-0.66

itation

-0.62

leasing

-0.60

 sqor

-0.60

pic

-0.60

 buddies

-0.60

McL

-0.59

POSITIVE LOGITS

etheus

0.93

ħĭ

0.90

ĪĴ

0.86

cases

0.81

furt

0.75

¥µ

0.74

Topic

0.71

Reason

0.70

 deserts

0.69

rium

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.