INDEX

Explanations

abstract concepts or words usually related to programming or academic contexts

oai_token-act-pair · gpt-3.5-turbo

concepts related to abstraction in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

odder

-0.76

UNCH

-0.72

ICAN

-0.70

cture

-0.69

wreck

-0.67

cano

-0.67

kins

-0.65

CVE

-0.64

udder

-0.63

risome

-0.62

POSITIVE LOGITS

ions

1.27

edly

0.88

urally

0.88

 algebra

0.87

tions

0.83

 syntax

0.81

 Matter

0.81

matter

0.80

 painter

0.79

ed

0.78

Activations Density 0.039%

abstract concepts or words usually related to programming or academic contexts

concepts related to abstraction in various contexts

No Comments

No Known Activations