INDEX

Explanations

references to sports scores or standings

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "the" and its relation to rankings and positions

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

atos

-0.63

anism

-0.63

 remembers

-0.62

amac

-0.60

forces

-0.59

ctors

-0.56

river

-0.56

 distinguishes

-0.54

 wallpaper

-0.54

 redistributed

-0.54

POSITIVE LOGITS

 expense

1.07

 outset

1.03

end

0.95

 helm

0.94

 behest

0.90

 conclusion

0.89

 halfway

0.83

 beginning

0.81

 start

0.79

 seams

0.79

Activations Density 0.078%

references to sports scores or standings

occurrences of the word "the" and its relation to rankings and positions

No Comments

No Known Activations