INDEX

Explanations

mentions of people with titles like "Rev." or "Dr." in the context of various subjects

oai_token-act-pair · gpt-3.5-turbo

references to religious figures or titles

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

etheless

-0.89

 snowball

-0.80

anwhile

-0.71

 overtake

-0.70

 feasible

-0.70

 accelerated

-0.69

 manoeuv

-0.69

 clutch

-0.69

lessly

-0.69

 leap

-0.68

POSITIVE LOGITS

 Kenneth

1.27

 Timothy

1.26

 Jeffrey

1.23

 Suzanne

1.20

 William

1.20

 Robert

1.19

 Marvin

1.19

 Jonathan

1.18

 Larry

1.18

 Michael

1.17

Activations Density 0.069%

mentions of people with titles like "Rev." or "Dr." in the context of various subjects

references to religious figures or titles

No Comments

No Known Activations