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

ufact

-0.76

iments

-0.71

itions

-0.68

agles

-0.67

icularly

-0.66

rix

-0.65

rys

-0.64

hers

-0.64

iple

-0.64

 backbone

-0.64

POSITIVE LOGITS

Death

0.78

 death

0.78

inventoryQuantity

0.76

 Death

0.76

death

0.69

 Embassy

0.68

 Survivors

0.68

 depreciation

0.68

 homophobia

0.66

 redemption

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.