INDEX

Explanations

phrases related to investigative or military activities

oai_token-act-pair · gpt-3.5-turbo

references to law enforcement or incidents related to authority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Same

-0.83

Same

-0.78

 Similar

-0.74

quickShipAvailable

-0.70

glers

-0.70

oother

-0.67

helps

-0.67

TOD

-0.67

 Variant

-0.66

 Lots

-0.63

POSITIVE LOGITS

 utterly

0.94

iless

0.86

 scarcely

0.85

 unparalleled

0.83

 brilliantly

0.83

 hitherto

0.81

 effortlessly

0.80

 decisively

0.79

 forty

0.79

 Auschwitz

0.79

Activations Density 1.317%

phrases related to investigative or military activities

references to law enforcement or incidents related to authority

No Comments

No Known Activations