INDEX

Explanations

phrases related to criminal activities and investigations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating ambiguity or uncertainty

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

hovah

-0.60

awa

-0.56

onto

-0.52

onga

-0.52

enegger

-0.50

emale

-0.50

ãĥ´

-0.48

ĪĴ

-0.48

 expires

-0.48

allas

-0.48

POSITIVE LOGITS

 yesterday

0.64

 Yanuk

0.54

 last

0.52

 earlier

0.51

 Priebus

0.51

 Arkham

0.48

 overnight

0.44

 Morsi

0.44

SELECT

0.43

 originally

0.42

Activations Density 2.596%

phrases related to criminal activities and investigations

phrases indicating ambiguity or uncertainty

No Comments

No Known Activations