INDEX

Explanations

references to "New York" in the text

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 retard

-0.78

rils

-0.74

 malt

-0.70

haar

-0.69

uler

-0.67

kin

-0.67

 borne

-0.66

bands

-0.65

azar

-0.64

claw

-0.64

POSITIVE LOGITS

New

3.24

New

2.56

NEW

1.94

NY

1.87

NYC

1.76

 Manhattan

1.71

 Newark

1.62

NEW

1.59

 Staten

1.56

 Brooklyn

1.51

Activations Density 0.046%

references to "New York" in the text

No Comments

No Known Activations