INDEX

Explanations

references to the term "Ho", particularly instances where it has a high activation value of 8 or 9

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the term "Ho," which is likely related to an organization, location, or significant recurrent theme in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 mileage

-0.73

 totality

-0.72

HIT

-0.66

CPC

-0.64

cape

-0.62

 mete

-0.62

CTR

-0.62

 nucleus

-0.62

 Mayhem

-0.61

 simultane

-0.61

POSITIVE LOGITS

pper

1.38

ppy

1.29

pping

1.23

oper

1.19

arding

1.15

oney

1.13

ppers

1.13

pped

1.11

ppe

1.10

ogle

1.09

Activations Density 0.020%

references to the term "Ho", particularly instances where it has a high activation value of 8 or 9

repeated mentions of the term "Ho," which is likely related to an organization, location, or significant recurrent theme in the text

No Comments

No Known Activations