INDEX

Explanations

short phrases of reported speech or quotes

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate attribution or citation of statements

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

ļéĨĴ

-1.00

¬¼

-0.94

displayText

-0.86

»Ĵ

-0.83

ģ«

-0.83

ĻĤ

-0.82

£ı

-0.79

ħĭ

-0.79

etheless

-0.78

ĪĴ

-0.74

POSITIVE LOGITS

 explains

1.67

 says

1.62

 admits

1.34

 explained

1.32

 observes

1.30

 recalls

1.25

 concedes

1.24

 warns

1.21

 adds

1.20

 argues

1.18

Activations Density 0.135%

short phrases of reported speech or quotes

phrases that indicate attribution or citation of statements

No Comments

No Known Activations