INDEX

Explanations

words related to legal matters and government departments, particularly the "Justice Department"

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

Not in Any Lists

No Comments

Negative Logits

ength

-0.72

-0.70

 livest

-0.69

 captcha

-0.68

ecd

-0.67

 resusc

-0.66

iple

-0.65

vre

-0.65

irst

-0.64

ontent

-0.63

POSITIVE LOGITS

 Department

1.12

 Dept

1.07

 League

0.92

 Scalia

0.89

 Assistance

0.82

 Gins

0.78

 Corps

0.78

 Anton

0.77

 Bureau

0.74

 Secretary

0.73

Activations Density 0.025%

words related to legal matters and government departments, particularly the "Justice Department"

No Comments

No Known Activations