INDEX

Explanations

articles (a, an) followed by a single word

oai_token-act-pair · gpt-3.5-turbo

articles and indefinite articles preceding nouns in various contexts

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

izons

-0.74

weights

-0.69

arsity

-0.68

hire

-0.67

fav

-0.65

enance

-0.65

usercontent

-0.64

llers

-0.64

itiz

-0.63

antes

-0.62

POSITIVE LOGITS

 nutshell

1.38

 statement

1.07

 flurry

1.00

 footnote

0.96

 brief

0.89

 twist

0.88

 keynote

0.87

 memorandum

0.86

 letter

0.86

 scathing

0.85

Activations Density 0.079%

articles (a, an) followed by a single word

articles and indefinite articles preceding nouns in various contexts

No Comments

No Known Activations