INDEX

Explanations

phrases related to interactions or relationships between different entities

oai_token-act-pair · gpt-3.5-turbo

instances of the word "and" associated with various subjects or relationships

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

YD

-0.79

needed

-0.73

Ĥİ

-0.72

oche

-0.70

wcs

-0.69

NEY

-0.67

INESS

-0.67

eker

-0.66

uit

-0.65

bard

-0.64

POSITIVE LOGITS

 halves

0.72

 nurture

0.68

EStreamFrame

0.67

 grasp

0.66

 à¨

0.65

 thence

0.64

 subsistence

0.64

 destiny

0.63

ingo

0.62

 vagina

0.62

Activations Density 0.097%

phrases related to interactions or relationships between different entities

instances of the word "and" associated with various subjects or relationships

No Comments

No Known Activations