INDEX

Explanations

mentions of dangerous or potentially lethal situations or substances

oai_token-act-pair · gpt-3.5-turbo

references to lethal situations or substances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ourced

-0.88

erous

-0.83

ership

-0.82

adra

-0.81

arity

-0.79

agate

-0.77

 CLIENT

-0.76

acular

-0.75

 Username

-0.75

orers

-0.74

POSITIVE LOGITS

 poisonous

0.91

 poison

0.91

 wounding

0.88

 dose

0.88

 assault

0.87

 lethal

0.85

 deadly

0.81

 weapon

0.81

ãĥ¥

0.81

 overdose

0.78

Activations Density 0.014%

mentions of dangerous or potentially lethal situations or substances

references to lethal situations or substances

No Comments

No Known Activations