INDEX

Explanations

locations, governmental bodies, and military terms

oai_token-act-pair · gpt-3.5-turbo

references to countries and their governments in a geopolitical context

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Interstitial

-0.84

Translation

-0.67

tnc

-0.66

omething

-0.63

Reading

-0.63

 sqor

-0.63

bm

-0.61

ammy

-0.61

table

-0.60

akings

-0.59

POSITIVE LOGITS

 denies

1.60

 accuses

1.43

 blames

1.41

 denounced

1.29

 blamed

1.28

 condemned

1.27

 insists

1.24

 denied

1.23

 vehemently

1.18

 condemns

1.16

Activations Density 0.247%

locations, governmental bodies, and military terms

references to countries and their governments in a geopolitical context

No Comments

No Known Activations