INDEX

Explanations

proper nouns, specifically names beginning with 'Ar'

oai_token-act-pair · gpt-3.5-turbo

references to the name "Ar."

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

 sight

-0.72

 birth

-0.63

 firsthand

-0.62

 functioning

-0.62

 reinvent

-0.62

 prominently

-0.58

 tested

-0.58

 clearly

-0.58

eners

-0.57

 sterling

-0.56

POSITIVE LOGITS

thritis

1.24

ranged

1.21

vind

1.21

ithmetic

1.20

lene

1.19

gue

1.16

agon

1.12

onson

1.11

rang

1.07

jun

1.07

Activations Density 0.013%

proper nouns, specifically names beginning with 'Ar'

references to the name "Ar."

No Comments

No Known Activations