INDEX

Explanations

proper nouns, specifically names like "Burton" and "Anchorage."

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific names and references related to individuals and organizations

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

atan

-0.88

payers

-0.79

olver

-0.76

ili

-0.74

fty

-0.72

ctions

-0.70

pace

-0.69

ivot

-0.68

tips

-0.68

endar

-0.67

POSITIVE LOGITS

 studios

0.68

NESS

0.67

inez

0.65

 Schwarzenegger

0.63

ning

0.62

lisher

0.59

anium

0.59

heimer

0.58

 Interstellar

0.58

unci

0.57

Activations Density 0.030%

proper nouns, specifically names like "Burton" and "Anchorage."

occurrences of specific names and references related to individuals and organizations

No Comments

No Known Activations