INDEX

Explanations

proper nouns and terms related to conspiracy theories, such as "Kendricks" and "tin-foil-hat-wearing conspiracy nuts"

oai_token-act-pair · gpt-3.5-turbo

instances of counting or measurement-related expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

abase

-0.73

 thous

-0.67

@#&

-0.65

MpServer

-0.64

[];

-0.63

 ultraviolet

-0.63

 remem

-0.62

 symp

-0.62

azeera

-0.62

undy

-0.61

POSITIVE LOGITS

bos

0.76

esty

0.74

los

0.67

bilt

0.65

liness

0.65

ocratic

0.64

operation

0.64

×Ļ×

0.63

oric

0.62

antics

0.61

Activations Density 0.000%

proper nouns and terms related to conspiracy theories, such as "Kendricks" and "tin-foil-hat-wearing conspiracy nuts"

instances of counting or measurement-related expressions

No Comments

No Known Activations