INDEX

Explanations

instances of someone being found guilty or accused of wrongdoing

oai_token-act-pair · gpt-3.5-turbo

phrases indicating legal findings or judgments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

lishes

-0.79

ibaba

-0.75

eatures

-0.73

yip

-0.73

enture

-0.70

dayName

-0.69

rium

-0.68

ription

-0.66

eeper

-0.65

 Flavoring

-0.64

POSITIVE LOGITS

 guilty

1.60

 ineligible

1.31

 unfit

1.27

 Guilty

1.24

 liable

1.20

 incapable

1.20

 incompetent

1.19

 unable

1.08

 culp

1.07

 innocent

1.05

Activations Density 0.094%

instances of someone being found guilty or accused of wrongdoing

phrases indicating legal findings or judgments

No Comments

No Known Activations