INDEX

Explanations

mentions of specific company names

oai_token-act-pair · gpt-3.5-turbo

mentions of companies

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

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

 survival

-0.69

 stab

-0.67

 swing

-0.67

 math

-0.65

 wand

-0.65

 magnet

-0.65

 violent

-0.64

 trunk

-0.63

 upset

-0.61

 overflow

-0.61

POSITIVE LOGITS

 Companies

4.07

Companies

3.08

 companies

1.97

Company

1.92

 Company

1.83

company

1.67

 Corpor

1.65

 Investors

1.56

 Brands

1.56

compan

1.55

Activations Density 0.013%

mentions of specific company names

mentions of companies

No Comments

No Known Activations