INDEX

Explanations

terms related to news agencies or journalism

oai_token-act-pair · gpt-3.5-turbo

mentions of the Associated Press (AP)

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 capitalist

-0.73

 stabil

-0.71

wagen

-0.69

lords

-0.68

 thumbs

-0.64

ãĤ´ãĥ³

-0.63

Reloaded

-0.63

 afore

-0.63

 capitalists

-0.63

abiding

-0.62

POSITIVE LOGITS

PLIED

1.18

PLE

1.06

PLIC

0.99

AP

0.99

sburg

0.95

TN

0.94

PAR

0.94

PRO

0.94

PE

0.92

PROV

0.92

Activations Density 0.010%

terms related to news agencies or journalism

mentions of the Associated Press (AP)

No Comments

No Known Activations