INDEX

Explanations

keywords related to cautioning against potential negative outcomes or actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lest" and words related to prohibition or caution

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

 Congratulations

-0.77

 Offline

-0.71

essor

-0.67

issance

-0.66

hardt

-0.65

estones

-0.64

Works

-0.64

 Thanks

-0.63

stood

-0.63

ulative

-0.63

POSITIVE LOGITS

 lest

1.09

entimes

0.81

 conce

0.79

 incur

0.79

 detract

0.78

fy

0.76

00200000

0.75

 overe

0.74

 fate

0.73

 fret

0.72

Activations Density 0.007%

keywords related to cautioning against potential negative outcomes or actions

instances of the word "lest" and words related to prohibition or caution

No Comments

No Known Activations