INDEX

Explanations

groups of people involving friends, family, and colleagues

oai_token-act-pair · gpt-3.5-turbo

references to groups or multiple individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 coffers

-0.71

 waterways

-0.70

-0.68

 bridges

-0.67

 stadiums

-0.67

 comparisons

-0.66

 ecosystems

-0.65

eries

-0.65

 Compatibility

-0.64

 defund

-0.64

POSITIVE LOGITS

 whom

0.88

 colleagues

0.85

brother

0.85

 coworkers

0.83

 colleague

0.83

 Malf

0.81

 comrades

0.81

who

0.79

who

0.78

 boyfriend

0.78

Activations Density 0.777%

groups of people involving friends, family, and colleagues

references to groups or multiple individuals

No Comments

No Known Activations