INDEX

Explanations

media and entertainment related terms, including TV, books, interviews, and reviews

oai_token-act-pair · gpt-3.5-turbo

references to events and activities related to media and entertainment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

cale

-0.78

fecture

-0.77

orah

-0.76

ressor

-0.74

uana

-0.72

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.72

rower

-0.71

 plaintiff

-0.70

 Located

-0.69

afia

-0.67

POSITIVE LOGITS

 announcements

1.60

 presentations

1.60

 speeches

1.54

 essays

1.45

 statements

1.42

 jokes

1.42

 videos

1.41

 anecdotes

1.39

 declarations

1.38

 stories

1.37

Activations Density 0.617%

media and entertainment related terms, including TV, books, interviews, and reviews

references to events and activities related to media and entertainment

No Comments

No Known Activations