INDEX

Explanations

the word "everything" used in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "everything."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

odan

-0.75

hement

-0.71

idon

-0.68

eme

-0.67

ename

-0.67

eka

-0.66

heed

-0.65

stood

-0.63

 lashes

-0.62

een

-0.62

POSITIVE LOGITS

 else

1.51

 imaginable

1.34

 Else

1.01

Else

0.98

 pertaining

0.95

 conceivable

0.93

 except

0.85

 happening

0.83

else

0.79

important

0.79

Activations Density 0.086%

the word "everything" used in various contexts

references to the concept of "everything."

No Comments

No Known Activations