INDEX

Explanations

phrases indicating reported information from various news sources

oai_token-act-pair · gpt-3.5-turbo

instances of news reporting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

isable

-0.78

anging

-0.73

otin

-0.71

enta

-0.71

ophers

-0.71

venge

-0.69

 gauge

-0.69

creen

-0.68

 palate

-0.68

osate

-0.66

POSITIVE LOGITS

 quoting

0.86

 sources

0.71

 citing

0.70

 yesterday

0.70

 excerpts

0.70

WASHINGTON

0.70

 ARTICLE

0.69

CBS

0.69

ILLE

0.68

biz

0.68

Activations Density 0.051%

phrases indicating reported information from various news sources

instances of news reporting

No Comments

No Known Activations