INDEX

Explanations

words related to specific geographical locations, particularly countries and cities

oai_token-act-pair · gpt-3.5-turbo

references to specific names and places, particularly focusing on individuals and geographical locations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

-1.03

iflower

-0.98

iard

-0.97

smith

-0.94

ibrary

-0.92

ipeg

-0.92

ski

-0.78

ileaks

-0.78

iosity

-0.78

ies

-0.77

POSITIVE LOGITS

chnology

0.98

 Centauri

0.92

OPLE

0.88

AKING

0.76

lda

0.76

xual

0.75

rosso

0.72

gger

0.71

 Gork

0.70

galitarian

0.69

Activations Density 0.153%

words related to specific geographical locations, particularly countries and cities

references to specific names and places, particularly focusing on individuals and geographical locations

No Comments

No Known Activations