INDEX

Explanations

positive descriptions of individuals, focusing on qualities such as being nice, good, and lovely

oai_token-act-pair · gpt-3.5-turbo

references to people, especially in a positive or descriptive context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Accessory

-0.80

uden

-0.75

DEN

-0.72

enz

-0.70

ROR

-0.70

eton

-0.70

Hart

-0.69

dos

-0.69

Stars

-0.69

Kar

-0.69

POSITIVE LOGITS

cule

0.68

 Olivia

0.68

nered

0.66

abul

0.64

 Malk

0.64

 foremost

0.64

ongevity

0.63

 proud

0.63

 nurt

0.62

onomic

0.60

Activations Density 0.188%

positive descriptions of individuals, focusing on qualities such as being nice, good, and lovely

references to people, especially in a positive or descriptive context

No Comments

No Known Activations