INDEX

Explanations

mentions of specific categories or classes

oai_token-act-pair · gpt-3.5-turbo

references to classification categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hyde

-0.90

oshenko

-0.77

mone

-0.69

aye

-0.68

Ó

-0.68

meier

-0.67

mos

-0.66

 reconciliation

-0.65

push

-0.63

 Cary

-0.62

POSITIVE LOGITS

 Class

3.87

Class

2.77

 class

2.49

 Classes

2.48

 CLASS

2.37

class

2.36

 classes

2.03

CLASS

1.87

classes

1.74

 Grade

1.59

Activations Density 0.012%

mentions of specific categories or classes

references to classification categories

No Comments

No Known Activations