INDEX

Explanations

references to a specific political news outlet

oai_token-act-pair · gpt-3.5-turbo

mentions of the news outlet Politico

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

lights

-0.79

ysis

-0.77

isted

-0.76

WWF

-0.76

oji

-0.68

cknowled

-0.64

ighters

-0.64

igned

-0.64

ettel

-0.64

umar

-0.64

POSITIVE LOGITS

 Caucus

0.93

 Magazine

0.90

 Digest

0.78

 aide

0.76

 Heights

0.75

 Insider

0.75

 Tanz

0.73

 Kills

0.73

icago

0.71

clinton

0.71

Activations Density 0.006%

references to a specific political news outlet

mentions of the news outlet Politico

No Comments

No Known Activations