INDEX

Explanations

keywords related to governmental entities, specifically the State Department

oai_token-act-pair · gpt-3.5-turbo

mentions of the State Department

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 begg

-0.62

irsch

-0.61

mos

-0.61

 belts

-0.61

 bour

-0.61

terday

-0.61

orde

-0.61

enegger

-0.60

 induct

-0.60

 pall

-0.59

POSITIVE LOGITS

 Department

1.38

 Dept

1.28

Department

0.93

Imp

0.89

 Secrets

0.85

house

0.85

 Farm

0.83

 Patrol

0.82

 Secretary

0.80

 Bureau

0.79

Activations Density 0.030%

keywords related to governmental entities, specifically the State Department

mentions of the State Department

No Comments

No Known Activations