INDEX

Explanations

phrases indicating permission or instruction

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "let."

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

ious

-0.79

stadt

-0.72

opal

-0.70

antz

-0.66

amer

-0.63

mans

-0.61

ipal

-0.60

omes

-0.59

ouses

-0.59

man

-0.55

POSITIVE LOGITS

let

3.35

 lets

2.00

Let

1.75

Let

1.68

 letting

1.61

 allow

1.57

LET

1.51

let

1.49

 Lets

1.26

 forgive

1.21

Activations Density 0.024%

phrases indicating permission or instruction

occurrences of the word "let."

No Comments

No Known Activations