INDEX

Explanations

law enforcement related terms and actions

oai_token-act-pair · gpt-3.5-turbo

references to law enforcement and official investigations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ãĥĸ

-0.75

 Dalai

-0.73

liest

-0.71

gets

-0.70

ABLE

-0.69

cast

-0.66

ocratic

-0.65

ãĥª

-0.65

Translation

-0.65

series

-0.65

POSITIVE LOGITS

 allege

0.98

 speculate

0.97

 confir

0.95

 raided

0.93

 encount

0.91

 investigating

0.86

 began

0.84

 believe

0.84

 concluded

0.84

 contend

0.83

Activations Density 0.150%

law enforcement related terms and actions

references to law enforcement and official investigations

No Comments

No Known Activations