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

lund

-0.65

please

-0.65

)",

-0.64

):

-0.63

—-

-0.60

olls

-0.58

åı

-0.58

âĢ¦."

-0.58

)"

-0.58

see

-0.58

POSITIVE LOGITS

imer

0.76

pol

0.76

lda

0.74

natureconservancy

0.73

aceae

0.72

Sap

0.72

DNA

0.70

arnaev

0.66

uno

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.