INDEX

Explanations

phrases indicating a strong emphasis on certain actions or situations

oai_token-act-pair · gpt-3.5-turbo

definite articles and contextually important nouns related to subjects and topics

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

DonaldTrump

-0.79

Zone

-0.75

Ĥ¬

-0.71

adelphia

-0.70

ZZ

-0.70

aeper

-0.68

HCR

-0.67

igslist

-0.67

PokÃ©

-0.65

izon

-0.65

POSITIVE LOGITS

 consequently

1.11

 accordingly

0.90

 consequ

0.87

 therefore

0.87

 alike

0.80

 resultant

0.77

 sooner

0.76

 subsequent

0.73

 moreover

0.68

 stakes

0.68

Activations Density 0.542%

phrases indicating a strong emphasis on certain actions or situations

definite articles and contextually important nouns related to subjects and topics

No Comments

No Known Activations