INDEX

Explanations

proper nouns, specifically potentially related to geographical locations or names of individuals

oai_token-act-pair · gpt-3.5-turbo

words and abbreviations related to locations, particularly names and terms associated with places

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ĸļ

-0.84

rogens

-0.63

addons

-0.61

 Uriel

-0.61

 caucuses

-0.58

ogun

-0.58

 bitter

-0.58

packages

-0.56

 lockout

-0.55

orem

-0.54

POSITIVE LOGITS

zinski

1.15

swick

0.99

sels

0.95

Ã§a

0.86

itte

0.82

combe

0.81

olini

0.81

wine

0.81

illet

0.80

cially

0.79

Activations Density 0.073%

proper nouns, specifically potentially related to geographical locations or names of individuals

words and abbreviations related to locations, particularly names and terms associated with places

No Comments

No Known Activations