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

tz

-0.80

stadt

-0.76

 Gingrich

-0.76

ice

-0.74

zers

-0.74

 ende

-0.71

vin

-0.68

 Wasserman

-0.66

 Cummings

-0.65

rir

-0.64

POSITIVE LOGITS

catentry

0.82

Surv

0.75

advertisement

0.74

Narr

0.70

POST

0.68

mere

0.68

)</

0.66

reet

0.64

Medium

0.63

rehens

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.