INDEX

Explanations

the word "yes" with varying activations, potentially in different contexts or expressions

oai_token-act-pair · gpt-3.5-turbo

affirmations or confirmations in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ember

-0.68

teenth

-0.66

agers

-0.66

Merit

-0.62

abo

-0.60

eland

-0.60

aching

-0.58

amorph

-0.57

Mobil

-0.56

POSITIVE LOGITS

yes

3.88

yes

2.28

YES

2.25

 yeah

2.13

Yes

2.13

Yes

1.98

YES

1.79

yea

1.59

yeah

1.35

 hello

1.27

Activations Density 0.011%

the word "yes" with varying activations, potentially in different contexts or expressions

affirmations or confirmations in the text

No Comments

No Known Activations