INDEX

Explanations

mentions of harm done to civilians in conflict situations

oai_token-act-pair · gpt-3.5-turbo

references to civilians in the context of conflict or war situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ologies

-0.81

CV

-0.80

Stud

-0.78

Lear

-0.75

Smith

-0.74

Hop

-0.73

ologically

-0.73

ENCE

-0.73

href

-0.72

ology

-0.72

POSITIVE LOGITS

 hostages

0.92

 casualties

0.90

 fleeing

0.88

 civilians

0.86

 slain

0.83

 evacuated

0.82

 massac

0.81

 displaced

0.80

 bystanders

0.79

 disarm

0.78

Activations Density 0.013%

mentions of harm done to civilians in conflict situations

references to civilians in the context of conflict or war situations

No Comments

No Known Activations