INDEX

Explanations

phrases related to warnings or notifications

oai_token-act-pair · gpt-3.5-turbo

instances of the word "alert."

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

Klu

-0.72

 Cheong

-0.71

esan

-0.70

Ì¶

-0.68

lot

-0.66

 phil

-0.64

vg

-0.63

VG

-0.62

 vend

-0.62

ogyn

-0.62

POSITIVE LOGITS

 alert

4.02

alert

2.80

 Alert

2.51

 alerts

2.35

Alert

2.00

 alerted

1.78

 alarm

1.55

 warning

1.48

 alarms

1.38

 warn

1.36

Activations Density 0.009%

phrases related to warnings or notifications

instances of the word "alert."

No Comments

No Known Activations