INDEX

Explanations

instances where actions are executed without authorization or consent

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "without" indicating actions or situations occurring in the absence of something else

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

oola

-0.77

ulty

-0.76

lish

-0.74

geist

-0.73

=-=-=-=-=-=-=-=-

-0.72

alyst

-0.70

 Inquisition

-0.70

des

-0.68

late

-0.66

raq

-0.66

POSITIVE LOGITS

 regard

0.95

lessly

0.91

 compromising

0.83

 knowing

0.80

 mentioning

0.79

 sacrificing

0.79

Charges

0.74

 necessarily

0.74

 interruption

0.73

 touching

0.73

Activations Density 0.038%

instances where actions are executed without authorization or consent

instances of the phrase "without" indicating actions or situations occurring in the absence of something else

No Comments

No Known Activations