INDEX

Explanations

punctuation marks, specifically periods

oai_token-act-pair · gpt-3.5-turbo

end-of-sentence punctuation

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

 predec

-0.77

 designated

-0.73

 permitted

-0.67

 advoc

-0.65

 contested

-0.65

 credential

-0.65

 authorized

-0.64

emaker

-0.64

 commissioned

-0.63

 eligible

-0.63

POSITIVE LOGITS

;)

1.11

:)

1.10

 Especially

1.02

 Anyway

1.00

:(

0.94

 Lets

0.94

:-)

0.94

 Sorry

0.93

 Seems

0.92

 Maybe

0.92

Activations Density 0.371%

punctuation marks, specifically periods

end-of-sentence punctuation

No Comments

No Known Activations