INDEX

Explanations

instances of reference to additional items or concepts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "additional."

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

bane

-0.77

bows

-0.76

 Alone

-0.66

orious

-0.65

abad

-0.65

generic

-0.64

 Forget

-0.63

dog

-0.63

 Theodore

-0.63

ESE

-0.62

POSITIVE LOGITS

judicial

0.93

 complication

0.86

 impetus

0.85

 installments

0.83

 opportunities

0.80

 avenues

0.79

 exposure

0.79

elaide

0.79

 additions

0.79

 expense

0.78

Activations Density 0.019%

instances of reference to additional items or concepts

instances of the word "additional."

No Comments

No Known Activations