INDEX

Explanations

web-related identifiers like URLs and computer science terms

oai_token-act-pair · gpt-3.5-turbo

various forms of punctuation and symbols often associated with programming or code formatting

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

erer

-0.72

cone

-0.67

minster

-0.66

Topic

-0.64

eda

-0.61

 ported

-0.60

GBT

-0.60

udos

-0.60

ense

-0.58

hor

-0.58

POSITIVE LOGITS

ordable

0.79

oice

0.72

iewicz

0.71

ammu

0.69

 EntityItem

0.67

 Flavoring

0.66

lished

0.63

weeney

0.62

qqa

0.61

iability

0.61

Activations Density 0.047%

web-related identifiers like URLs and computer science terms

various forms of punctuation and symbols often associated with programming or code formatting

No Comments

No Known Activations