INDEX

Explanations

phrases related to events, possibly including the word "party."

oai_token-act-pair · gpt-3.5-turbo

references to political parties

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 composite

-0.70

eval

-0.66

bos

-0.64

sc

-0.63

nell

-0.63

 bible

-0.62

feature

-0.62

 android

-0.60

ĻĤ

-0.59

ones

-0.59

POSITIVE LOGITS

Party

3.54

 Party

3.35

 Parties

2.84

party

2.11

 party

2.08

 parties

1.59

 Kart

1.58

 Lobby

1.22

 Caucus

1.20

Political

1.18

Activations Density 0.033%

phrases related to events, possibly including the word "party."

references to political parties

No Comments

No Known Activations