INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

mentions of specific individuals, particularly "Wendy" and "Nad."

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

umen

-0.88

heed

-0.86

ration

-0.81

ular

-0.78

ulus

-0.76

rated

-0.75

lesi

-0.75

hement

-0.74

ports

-0.74

achev

-0.73

POSITIVE LOGITS

 Wendy

0.78

 Fitzgerald

0.73

 Stephenson

0.72

||||

0.70

 Rice

0.68

bourg

0.68

 Tire

0.67

 Morton

0.67

McN

0.66

 Past

0.66

Activations Density 0.034%

names of specific individuals

mentions of specific individuals, particularly "Wendy" and "Nad."

No Comments

No Known Activations