INDEX

Explanations

mentions of monetary penalties or fines

oai_token-act-pair · gpt-3.5-turbo

instances of fines being imposed

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

 Awakening

-0.72

 Abrams

-0.68

 Project

-0.65

 Advent

-0.64

 realization

-0.63

 projection

-0.62

 Ruby

-0.62

ancer

-0.62

 survival

-0.61

xes

-0.60

POSITIVE LOGITS

 fined

3.97

 fines

1.68

 punished

1.60

 sentenced

1.52

 prosecuted

1.32

 penal

1.28

fine

1.22

 sued

1.20

 rewarded

1.20

 indicted

1.19

Activations Density 0.013%

mentions of monetary penalties or fines

instances of fines being imposed

No Comments

No Known Activations