INDEX

Explanations

the word "why" as well as phrases emphasizing reasons or justifications

oai_token-act-pair · gpt-3.5-turbo

reasons or justifications for actions and concepts

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

ãĥ¼ãĤ¯

-0.86

sylv

-0.75

incial

-0.75

ãĥĥãĥĪ

-0.73

iox

-0.73

astical

-0.72

externalActionCode

-0.71

ortment

-0.70

illus

-0.69

duino

-0.68

POSITIVE LOGITS

we

1.04

 they

0.87

0.81

it

0.80

rely

0.79

why

0.75

you

0.71

soever

0.70

so

0.70

he

0.68

Activations Density 0.059%

the word "why" as well as phrases emphasizing reasons or justifications

reasons or justifications for actions and concepts

No Comments

No Known Activations