INDEX

Explanations

words related to disguises or pretending to be something else

oai_token-act-pair · gpt-3.5-turbo

terms related to masquerading or false identities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

NEC

-0.66

 Cosponsors

-0.65

STER

-0.63

RSS

-0.61

 STATS

-0.61

 DRAG

-0.60

 RELEASE

-0.59

 Schumer

-0.58

 socket

-0.57

 closing

-0.56

POSITIVE LOGITS

quer

1.25

ading

1.07

querade

0.98

ior

0.92

aders

0.89

iors

0.87

aque

0.86

ader

0.84

aban

0.84

ÃŃa

0.83

Activations Density 0.010%

words related to disguises or pretending to be something else

terms related to masquerading or false identities

No Comments

No Known Activations