INDEX

Explanations

proper nouns or names of people and places

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

•ans

#4 activation: 9.58

No Comments

Negative Logits

-+-+

-0.76

 Kinnikuman

-0.76

ĳå£«

-0.68

ORPG

-0.68

=-=-=-=-

-0.65

OHN

-0.62

 informational

-0.60

²¾

-0.59

Ħ¢

-0.59

 Sending

-0.58

POSITIVE LOGITS

ibaba

1.04

bucks

1.00

iland

0.99

lde

0.85

buck

0.81

risome

0.74

ulkan

0.73

kees

0.72

erville

0.70

tering

0.70

Activations Density 0.048%

proper nouns or names of people and places

No Comments

No Known Activations