INDEX

Explanations

references to statistical information or reports

oai_token-act-pair · gpt-3.5-turbo

instances of people or groups referring to themselves in plural forms

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

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

 Rhodes

-0.83

 Rack

-0.73

 Zeit

-0.71

 Linear

-0.70

 Tulsa

-0.69

 Hilton

-0.68

 JPEG

-0.67

 Brass

-0.67

 Mirage

-0.67

 Catalyst

-0.66

POSITIVE LOGITS

ï¸ı

1.23

who

1.11

selves

1.00

parents

0.92

owners

0.91

resent

0.91

ividual

0.89

eligible

0.88

Â£

0.87

whose

0.87

Activations Density 0.193%

references to statistical information or reports

instances of people or groups referring to themselves in plural forms

No Comments

No Known Activations