INDEX

Explanations

mentions of military divisions and combat activities

oai_token-act-pair · gpt-3.5-turbo

references to military divisions and operations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Property

-0.81

Consumer

-0.80

 pores

-0.79

-->

-0.75

ï¸ı

-0.74

Gi

-0.71

ðĿ

-0.70

à

-0.70

FACE

-0.69

Energy

-0.69

POSITIVE LOGITS

 colonel

1.20

utenant

1.12

 stationed

1.08

alions

1.08

 commander

1.07

 commanders

1.05

geant

1.03

selage

1.02

alion

1.02

 troops

1.01

Activations Density 0.643%

mentions of military divisions and combat activities

references to military divisions and operations

No Comments

No Known Activations