INDEX

Explanations

phrases indicating alignment or support towards a particular side or position in a debate or conflict

oai_token-act-pair · gpt-3.5-turbo

references to different sides in arguments or disputes, particularly in a political context

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

lear

-0.86

frey

-0.82

ardy

-0.78

rss

-0.76

enegger

-0.73

htaking

-0.73

encers

-0.71

yssey

-0.71

untu

-0.70

imens

-0.70

POSITIVE LOGITS

kick

0.92

 seams

0.83

 seam

0.77

Za

0.76

side

0.76

 sidelines

0.73

boards

0.70

 side

0.70

board

0.70

 altar

0.68

Activations Density 0.021%

phrases indicating alignment or support towards a particular side or position in a debate or conflict

references to different sides in arguments or disputes, particularly in a political context

No Comments

No Known Activations