INDEX

Explanations

countries and political terms related to political systems

oai_token-act-pair · gpt-3.5-turbo

references to conservative and nationalist political ideologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

anooga

-0.74

nesota

-0.74

Todd

-0.72

Avg

-0.68

VERS

-0.68

addon

-0.68

ebus

-0.67

 Overs

-0.67

Crash

-0.65

Wow

-0.65

POSITIVE LOGITS

 Sharia

1.28

 citizenship

1.23

 parliamentary

1.20

 passports

1.18

 parliament

1.16

 nationality

1.16

 decriminal

1.14

 amnesty

1.13

 compulsory

1.10

 dictatorship

1.09

Activations Density 0.732%

countries and political terms related to political systems

references to conservative and nationalist political ideologies

No Comments

No Known Activations