INDEX

Explanations

references related to conflict, contentious incidents, or human rights violations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating actions or violations related to invasion and oppression

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

 partName

-0.82

 istg

-0.71

 ancest

-0.71

cano

-0.65

soType

-0.64

bet

-0.64

ringe

-0.64

aldo

-0.61

eus

-0.61

fters

-0.60

POSITIVE LOGITS

 sorts

0.76

 fossil

0.68

 Goods

0.67

 course

0.67

 Trafford

0.66

 civilians

0.64

 wildlife

0.63

ARS

0.62

 minorities

0.59

 Warfare

0.59

Activations Density 0.169%

references related to conflict, contentious incidents, or human rights violations

phrases indicating actions or violations related to invasion and oppression

No Comments

No Known Activations