INDEX

Explanations

words related to hierarchical structures, especially rankings and positions within an organization or society

oai_token-act-pair · gpt-3.5-turbo

references to hierarchical positions or levels of authority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

vous

-0.91

argo

-0.74

 Sparkle

-0.70

 Apocalypse

-0.69

issan

-0.69

 Archangel

-0.69

issance

-0.68

 Ezek

-0.66

 Sunset

-0.65

reb

-0.65

POSITIVE LOGITS

 rank

1.34

Rank

1.25

 ranks

0.98

 Rank

0.95

rank

0.95

 rankings

0.92

 ranking

0.88

 ranked

0.86

ranking

0.84

 insign

0.83

Activations Density 0.004%

words related to hierarchical structures, especially rankings and positions within an organization or society

references to hierarchical positions or levels of authority

No Comments

No Known Activations