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

Scope

-0.79

ĺħ

-0.76

 faults

-0.65

ĵĺ

-0.65

 braces

-0.63

ortment

-0.63

 Seym

-0.62

 fault

-0.61

 encour

-0.60

isSpecial

-0.60

POSITIVE LOGITS

Downloadha

0.86

ickle

0.85

peak

0.85

hyper

0.82

quickShipAvailable

0.77

APD

0.76

Hyper

0.73

OH

0.72

hooting

0.72

zin

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.