INDEX

Explanations

phrases related to expressing opinions or positions

oai_token-act-pair · gpt-3.5-turbo

end punctuation marks, particularly periods, in statements conveying certainty

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ãĥĻ

-0.66

 jugg

-0.66

 conceal

-0.62

utical

-0.62

ãĤ©

-0.62

 prey

-0.61

 steadily

-0.61

 disappear

-0.61

 sneak

-0.61

 curtain

-0.60

POSITIVE LOGITS

 Specifically

0.87

Writing

0.87

``

0.86

 Saying

0.75

 According

0.75

 Asked

0.74

"[

0.74

 However

0.73

 Called

0.72

 Speaking

0.72

Activations Density 0.425%

phrases related to expressing opinions or positions

end punctuation marks, particularly periods, in statements conveying certainty

No Comments

No Known Activations