INDEX

Explanations

names of individuals or entities with the suffix "-ke"

oai_token-act-pair · gpt-3.5-turbo

proper nouns and specific references to people, places, or entities

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

»Ĵ

-0.66

ãĤ¨

-0.66

ĺ

-0.64

±

-0.63

-+-+

-0.61

Ķ

-0.60

 unden

-0.58

ccording

-0.58

 conflic

-0.57

cedented

-0.56

POSITIVE LOGITS

lords

0.68

sburg

0.68

lain

0.64

ebook

0.62

eers

0.61

 Crusade

0.60

 Handbook

0.60

tower

0.60

gery

0.59

hare

0.59

Activations Density 1.111%

names of individuals or entities with the suffix "-ke"

proper nouns and specific references to people, places, or entities

No Comments

No Known Activations

names of individuals or entities with the suffix "-ke"

proper nouns and specific references to people, places, or entities

No Comments

No Known Activations