INDEX

Explanations

words related to specific attributes or features of objects or individuals

oai_token-act-pair · gpt-3.5-turbo

references to defining traits or features of people, objects, or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

×ª

-0.72

×Ķ

-0.71

GS

-0.69

pg

-0.68

kos

-0.68

NAS

-0.67

ski

-0.66

POR

-0.65

gur

-0.65

×©

-0.64

POSITIVE LOGITS

 characteristics

1.28

istics

1.12

 traits

1.10

 charact

1.02

eatures

0.99

 qualities

0.96

 weaknesses

0.93

 similarities

0.88

 attributes

0.88

Features

0.87

Activations Density 0.011%

words related to specific attributes or features of objects or individuals

references to defining traits or features of people, objects, or concepts

No Comments

No Known Activations