INDEX

Explanations

words related to medical conditions, especially those ending in "-ap" such as "apologist" or "apathy"

oai_token-act-pair · gpt-3.5-turbo

references to "apologists" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Hurricanes

-0.80

fast

-0.66

fry

-0.65

 Instruments

-0.64

 naughty

-0.63

 Sands

-0.63

 Dynamics

-0.62

enegger

-0.61

 Dolphins

-0.61

 Carib

-0.61

POSITIVE LOGITS

athetic

1.42

artment

1.38

ocry

1.35

ocaly

1.34

ollo

1.30

athy

1.28

olitical

1.27

otrop

1.25

osite

1.22

ocalyptic

1.21

Activations Density 0.023%

words related to medical conditions, especially those ending in "-ap" such as "apologist" or "apathy"

references to "apologists" in various contexts

No Comments

No Known Activations