INDEX

Explanations

the word "Past"

oai_token-act-pair · gpt-3.5-turbo

references to historical contexts or past events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

anguage

-0.78

izes

-0.72

UGH

-0.70

 EntityItem

-0.70

oaded

-0.69

DERR

-0.65

jriwal

-0.65

ized

-0.64

ENGTH

-0.63

izable

-0.63

POSITIVE LOGITS

oral

1.26

ebin

0.98

fam

0.95

orius

0.90

urious

0.89

olini

0.88

ors

0.88

urity

0.87

ure

0.83

asus

0.80

Activations Density 0.028%

the word "Past"

references to historical contexts or past events

No Comments

No Known Activations