INDEX

Explanations

names of people, specifically with a focus on surnames

oai_token-act-pair · gpt-3.5-turbo

names of individuals associated with leadership or authority

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

TX

-0.69

cha

-0.66

 flares

-0.65

 Metatron

-0.65

 separ

-0.63

aea

-0.63

 travellers

-0.62

 blend

-0.62

 visc

-0.61

 curv

-0.61

POSITIVE LOGITS

Paw

3.94

paw

1.25

 Rasmussen

1.23

Yad

1.22

 Ramsey

1.14

zinski

1.12

alsh

1.08

Hew

1.01

ÅĤ

0.98

Pav

0.98

Activations Density 0.028%

names of people, specifically with a focus on surnames

names of individuals associated with leadership or authority

No Comments

No Known Activations