INDEX

Explanations

phrases related to police intervention or response

oai_token-act-pair · gpt-3.5-turbo

occurrences of police activity and response

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

cair

-0.68

 autobi

-0.68

 heir

-0.67

 maiden

-0.65

rawdownloadcloneembedreportprint

-0.64

 embodied

-0.64

Sov

-0.64

XY

-0.64

iku

-0.63

PRES

-0.63

POSITIVE LOGITS

 paramedics

0.97

 patrols

0.93

 Detect

0.90

 checkpoints

0.88

 patrolling

0.84

 lookout

0.83

 handc

0.83

 patrol

0.83

 arrests

0.82

nsics

0.82

Activations Density 0.447%

phrases related to police intervention or response

occurrences of police activity and response

No Comments

No Known Activations