INDEX

Explanations

mentions of law enforcement agencies or figures

oai_token-act-pair · gpt-3.5-turbo

mentions of authorities and their actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

thal

-0.78

leaf

-0.76

ces

-0.70

TON

-0.67

vious

-0.67

bow

-0.67

tom

-0.66

 Candle

-0.65

READ

-0.65

rous

-0.64

POSITIVE LOGITS

 authorities

1.11

 Authorities

0.94

 agencies

0.91

 confir

0.86

 officials

0.86

 raided

0.84

 resorted

0.82

 tasked

0.80

 Officials

0.78

 consulted

0.77

Activations Density 0.017%

mentions of law enforcement agencies or figures

mentions of authorities and their actions

No Comments

No Known Activations