INDEX

Explanations

mentions of groups and organizations, specifically those related to advocacy and rights

oai_token-act-pair · gpt-3.5-turbo

references to advocacy and environmental or human rights groups

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

lator

-0.72

colour

-0.69

ussen

-0.65

 unmarked

-0.64

depend

-0.64

 liqu

-0.64

ãĥİ

-0.63

ipment

-0.62

ãĤ¨ãĥ«

-0.62

 transporter

-0.61

POSITIVE LOGITS

 ACLU

0.76

 Farrell

0.75

 Beir

0.74

 nonprofits

0.74

 Parenthood

0.74

 attorney

0.73

 advocacy

0.72

 Feinstein

0.71

 sued

0.70

 groups

0.69

Activations Density 0.129%

mentions of groups and organizations, specifically those related to advocacy and rights

references to advocacy and environmental or human rights groups

No Comments

No Known Activations