INDEX

Explanations

nouns referring to persons or entities

oai_token-act-pair · gpt-3.5-turbo

references to authors or personalities mentioned in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Creed

-0.73

 Ghosts

-0.64

cy

-0.60

 mode

-0.60

jog

-0.60

 speed

-0.59

 strength

-0.59

 reflex

-0.59

 turbo

-0.58

sac

-0.58

POSITIVE LOGITS

eller

4.96

elling

2.03

ellar

1.79

ell

1.70

elled

1.59

ells

1.37

ellery

1.32

ellen

1.32

ella

1.26

eele

1.24

Activations Density 0.006%

nouns referring to persons or entities

references to authors or personalities mentioned in the text

No Comments

No Known Activations