INDEX

Explanations

words related to fire

oai_token-act-pair · gpt-3.5-turbo

references to fire, including its intensity and associated imagery

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

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

 Lans

-0.74

 Vide

-0.65

ortment

-0.63

 newsp

-0.63

IFIED

-0.62

edy

-0.62

 Freed

-0.60

eters

-0.59

 Citizen

-0.59

herty

-0.58

POSITIVE LOGITS

flies

1.39

storm

1.38

fly

1.27

balls

1.26

brand

1.26

 exting

1.17

proof

1.15

storms

1.15

bolt

1.11

bomb

1.10

Activations Density 0.043%

words related to fire

references to fire, including its intensity and associated imagery

No Comments

No Known Activations