INDEX

Explanations

news-related phrases and updates

oai_token-act-pair · gpt-3.5-turbo

updates and information related to current news and announcements

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

 Painter

-0.75

PLIED

-0.73

conservancy

-0.73

limits

-0.71

ħĭ

-0.66

aterasu

-0.66

krit

-0.64

ngth

-0.64

 blinded

-0.62

iors

-0.62

POSITIVE LOGITS

 news

1.98

 developments

1.76

 announcements

1.62

 announcement

1.48

 NEWS

1.46

news

1.39

 rumor

1.33

tid

1.30

 rumors

1.28

 Announce

1.26

Activations Density 0.798%

news-related phrases and updates

updates and information related to current news and announcements

No Comments

No Known Activations