INDEX

Explanations

information related to crime, justice, and political investigations

oai_token-act-pair · gpt-3.5-turbo

references to violent incidents and legal actions related to murder and political events

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

 denotes

-0.63

emonium

-0.59

staking

-0.58

 summed

-0.53

erd

-0.52

 anecd

-0.52

speak

-0.51

JV

-0.50

lished

-0.49

urai

-0.49

POSITIVE LOGITS

 illegally

0.60

AFP

0.58

 unfairly

0.54

 idols

0.53

 amid

0.53

 allegedly

0.52

 ObamaCare

0.52

 hostages

0.51

 detainees

0.49

 empire

0.49

Activations Density 1.921%

information related to crime, justice, and political investigations

references to violent incidents and legal actions related to murder and political events

No Comments

No Known Activations