INDEX

Explanations

news-related keywords

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "News" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

actory

-0.71

itent

-0.71

otent

-0.70

ibr

-0.70

resent

-0.70

ading

-0.69

 skilled

-0.68

ugh

-0.68

aded

-0.67

 goddamn

-0.67

POSITIVE LOGITS

News

1.25

NEWS

1.00

 Articles

0.92

 headlines

0.91

Content

0.88

 News

0.88

orial

0.85

 Coverage

0.85

 Seym

0.82

Reporting

0.82

Activations Density 0.009%

news-related keywords

occurrences of the word "News" in various contexts

No Comments

No Known Activations