INDEX

Explanations

specific names of organizations and people

oai_token-act-pair · gpt-3.5-turbo

references to specific brands, products, or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

)=

-0.69

 mattered

-0.68

 hump

-0.68

 defer

-0.65

 swallowed

-0.65

displayText

-0.63

 yawn

-0.62

 nonexistent

-0.62

 disqual

-0.62

inventoryQuantity

-0.61

POSITIVE LOGITS

Ltd

1.22

 Productions

1.15

LLP

1.10

 Labs

1.09

LLC

1.07

 Architects

1.06

 Forums

1.06

LTD

1.04

 Studios

0.99

 Limited

0.98

Activations Density 0.492%

specific names of organizations and people

references to specific brands, products, or organizations

No Comments

No Known Activations