INDEX

Explanations

phrases related to actions that have not been done or experienced yet

oai_token-act-pair · gpt-3.5-turbo

instances of the word "haven" in various forms

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

 princ

-0.84

aditional

-0.83

 eleph

-0.82

 newcom

-0.78

 unnecess

-0.77

 withd

-0.76

ò

-0.75

 challeng

-0.74

 pione

-0.73

ß

-0.71

POSITIVE LOGITS

't

1.60

dayName

0.91

Â´

0.89

ÃŃ

0.88

ited

0.87

izon

0.82

utm

0.82

seen

0.82

thouse

0.81

istar

0.81

Activations Density 0.040%

phrases related to actions that have not been done or experienced yet

instances of the word "haven" in various forms

No Comments

No Known Activations