INDEX

Explanations

references to geopolitical entities

oai_token-act-pair · gpt-3.5-turbo

mentions of Russian entities or related terms

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

phe

-0.96

wn

-0.78

ÅĤ

-0.76

now

-0.73

ivable

-0.73

illy

-0.72

track

-0.70

ÃŁ

-0.70

ifice

-0.70

reviewed

-0.70

POSITIVE LOGITS

 diplomats

1.18

 authorities

1.12

 nationals

1.11

 marines

1.08

 Foreign

1.08

 Ambassador

1.06

 officials

1.05

 Prime

1.04

 President

1.04

 troops

1.03

Activations Density 0.105%

references to geopolitical entities

mentions of Russian entities or related terms

No Comments

No Known Activations