INDEX

Explanations

Breaking news alerts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Breaking" related to news content

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

geries

-0.82

minist

-0.79

Reviewer

-0.74

iors

-0.71

geons

-0.70

ivan

-0.69

ogg

-0.69

uther

-0.67

geon

-0.66

ronic

-0.64

POSITIVE LOGITS

 News

1.12

 news

0.99

News

0.95

views

0.95

 NEWS

0.92

Bad

0.90

news

0.85

fast

0.80

NEWS

0.80

0.76

Activations Density 0.022%

Breaking news alerts

occurrences of the word "Breaking" related to news content

No Comments

No Known Activations