INDEX

Explanations

the word "to" along with a high activation value

oai_token-act-pair · gpt-3.5-turbo

expressions of difficulty or challenge

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

afety

-0.74

mails

-0.71

quickShipAvailable

-0.68

ONSORED

-0.65

listed

-0.64

Dash

-0.63

 packets

-0.62

 airs

-0.62

 coats

-0.60

Orig

-0.59

POSITIVE LOGITS

ggles

1.08

 avoid

1.08

 achieve

1.07

adies

1.05

 accomplish

1.04

 satisfy

1.04

 accommodate

1.04

 fulfill

1.01

 obtain

1.00

 keep

1.00

Activations Density 0.559%

the word "to" along with a high activation value

expressions of difficulty or challenge

No Comments

No Known Activations