INDEX

Explanations

words related to owing something, such as debts, promises, or gratitude

oai_token-act-pair · gpt-3.5-turbo

references to financial obligations or debts

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

 amplification

-0.65

activ

-0.62

 safety

-0.60

 scanner

-0.59

 narrow

-0.59

 Detection

-0.59

 mass

-0.58

 selection

-0.58

ference

-0.58

MK

-0.58

POSITIVE LOGITS

 owed

3.48

owe

3.40

 owes

3.31

 owing

1.66

 indebted

1.51

 debts

1.43

 repaid

1.29

 debt

1.20

 borrowed

1.15

 creditors

1.13

Activations Density 0.028%

words related to owing something, such as debts, promises, or gratitude

references to financial obligations or debts

No Comments

No Known Activations