INDEX

Explanations

the word "Me" with varying degrees of emphasis indicated by activation strength

oai_token-act-pair · gpt-3.5-turbo

the repetition of the word "Me"

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

UAL

-0.72

ctl

-0.66

icably

-0.65

flush

-0.65

acing

-0.64

 flush

-0.63

ulative

-0.62

ript

-0.62

itiveness

-0.62

OWER

-0.60

POSITIVE LOGITS

Me

3.48

Me

2.45

ME

2.04

me

1.77

Us

1.49

Meh

1.41

ME

1.36

Him

1.33

My

1.27

me

1.24

Activations Density 0.012%

the word "Me" with varying degrees of emphasis indicated by activation strength

the repetition of the word "Me"

No Comments

No Known Activations