INDEX

Explanations

phrases related to suspected individuals, particularly in the context of crime or wrongdoing

oai_token-act-pair · gpt-3.5-turbo

instances of the word "suspected" related to criminal activities or accusations

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

rx

-0.70

idge

-0.68

skill

-0.67

 hither

-0.66

learn

-0.64

ummer

-0.63

jri

-0.63

learning

-0.62

hire

-0.62

lot

-0.62

POSITIVE LOGITS

iary

0.74

 suspect

0.72

 arson

0.70

 culprit

0.69

 infring

0.65

lessly

0.64

 suspects

0.63

 mishand

0.63

 guilty

0.63

 vandal

0.62

Activations Density 0.018%

phrases related to suspected individuals, particularly in the context of crime or wrongdoing

instances of the word "suspected" related to criminal activities or accusations

No Comments

No Known Activations