INDEX

Explanations

news-related terms involving police activities and investigations

oai_token-act-pair · gpt-3.5-turbo

active sentences containing police-related actions or mentions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 unworthy

-0.73

 moderation

-0.73

 detrim

-0.69

 pinnacle

-0.67

 whining

-0.66

 embod

-0.65

opian

-0.65

 advoc

-0.65

ufact

-0.64

 Madness

-0.63

POSITIVE LOGITS

 arresting

1.33

 arrests

1.28

 Detect

1.21

 investigating

1.19

 arrest

1.16

 arrested

1.14

 apprehend

1.11

 raided

1.08

 detectives

1.08

Detect

1.07

Activations Density 0.347%

news-related terms involving police activities and investigations

active sentences containing police-related actions or mentions

No Comments

No Known Activations