INDEX

Explanations

references to physical harm or danger in a context of warnings or incidents

oai_token-act-pair · gpt-3.5-turbo

keywords related to injuries and safety concerns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ku

-0.86

liam

-0.75

kt

-0.74

ãĤ©

-0.72

atl

-0.71

utm

-0.68

kh

-0.68

last

-0.68

ãĤ¨

-0.66

achment

-0.64

POSITIVE LOGITS

nor

1.15

 anymore

1.06

 whatsoever

1.03

Initialized

0.73

 slightest

0.68

;}

0.68

 detectable

0.68

 anybody

0.67

 anything

0.66

uyomi

0.66

Activations Density 0.653%

references to physical harm or danger in a context of warnings or incidents

keywords related to injuries and safety concerns

No Comments

No Known Activations