INDEX

Explanations

phrases indicating something exists or is located in a certain place

oai_token-act-pair · gpt-3.5-turbo

references to the existence of various things or entities

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

CJ

-0.66

mut

-0.64

UGC

-0.63

ories

-0.60

 Carbuncle

-0.60

Ã§

-0.59

 Younger

-0.59

 Prince

-0.58

fully

-0.58

PART

-0.56

POSITIVE LOGITS

abouts

1.25

ngth

0.81

 flyers

0.79

 pregn

0.71

tics

0.68

dden

0.67

hole

0.67

holes

0.67

comings

0.65

rared

0.65

Activations Density 0.017%

phrases indicating something exists or is located in a certain place

references to the existence of various things or entities

No Comments

No Known Activations