INDEX

Explanations

terms related to news articles reporting violence and conflict

oai_token-act-pair · gpt-3.5-turbo

references to violence and terrorism

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 nutrit

-0.66

 cushion

-0.57

 availability

-0.57

 Refresh

-0.55

 nutritious

-0.54

pool

-0.54

cub

-0.54

ository

-0.53

 Canal

-0.53

 supplementary

-0.52

POSITIVE LOGITS

Äĩ

0.76

Ń·

0.73

 Boko

0.67

 Afgh

0.67

vic

0.65

Latin

0.63

ilan

0.63

 Hezbollah

0.63

vous

0.62

 Donetsk

0.62

Activations Density 1.258%

terms related to news articles reporting violence and conflict

references to violence and terrorism

No Comments

No Known Activations