INDEX

Explanations

proper nouns or names containing the letters "ko"

oai_token-act-pair · gpt-3.5-turbo

references to the name "Ko"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

glass

-0.76

Ø¨

-0.73

Ö¼

-0.72

 Creed

-0.71

ingham

-0.71

 narrator

-0.70

senal

-0.67

 à¨

-0.65

å¤©

-0.65

ibly

-0.63

POSITIVE LOGITS

zzi

1.21

pport

0.96

osta

0.95

essler

0.94

pper

0.93

jo

0.92

ppa

0.90

pps

0.90

zy

0.89

unin

0.89

Activations Density 0.025%

proper nouns or names containing the letters "ko"

references to the name "Ko"

No Comments

No Known Activations