INDEX

Explanations

news and military related words and phrases

oai_token-act-pair · gpt-3.5-turbo

references to military actions and statements

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

awei

-0.76

ı

-0.74

yt

-0.72

advertisement

-0.71

enh

-0.68

alog

-0.67

erson

-0.67

toe

-0.67

alf

-0.66

gdala

-0.65

POSITIVE LOGITS

 Nasa

0.83

 petrol

0.72

alysed

0.71

 Nato

0.68

 apologise

0.68

 Ukip

0.66

 signalling

0.65

cially

0.65

PID

0.65

 ageing

0.65

Activations Density 0.122%

news and military related words and phrases

references to military actions and statements

No Comments

No Known Activations