INDEX

Explanations

mentions of the number "two" or pairings

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

taboola

-0.73

 EVENTS

-0.65

pmwiki

-0.62

humane

-0.59

ople

-0.59

VERTISEMENT

-0.56

externalActionCode

-0.55

Fre

-0.55

EngineDebug

-0.54

Ĥİ

-0.54

POSITIVE LOGITS

 apiece

1.32

 totaling

1.12

 respectively

1.00

 simultaneously

0.99

 consecut

0.91

 vying

0.90

 separated

0.89

 concurrently

0.88

 together

0.86

 collided

0.86

Activations Density 0.447%

mentions of the number "two" or pairings

No Comments

No Known Activations