INDEX

Explanations

phrases related to medical emergencies

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 challeng

-0.97

 millenn

-0.83

 tempt

-0.82

 imagining

-0.81

 boycot

-0.81

 embodied

-0.79

inav

-0.79

 lett

-0.78

 enriched

-0.78

 creeping

-0.77

POSITIVE LOGITS

 Investigators

1.51

 Authorities

1.38

Investigators

1.30

 Anyone

1.29

 Officers

1.28

 Officials

1.26

Officers

1.24

 Witnesses

1.22

 Detect

1.22

Police

1.19

Activations Density 0.236%

phrases related to medical emergencies

No Comments

No Known Activations