INDEX

Explanations

phrases indicating giving something to someone or providing information

oai_token-act-pair · gpt-3.5-turbo

commands or requests directed at others

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

éĹ

-0.74

lit

-0.71

istration

-0.67

*/(

-0.67

cffffcc

-0.66

 phyl

-0.65

 discharge

-0.64

pher

-0.62

ļé

-0.61

 controlling

-0.61

POSITIVE LOGITS

 Your

0.98

 Yourself

0.91

 Them

0.88

Now

0.84

 Help

0.80

 Take

0.80

 Injury

0.78

Take

0.77

Leave

0.76

 Give

0.76

Activations Density 0.111%

phrases indicating giving something to someone or providing information

commands or requests directed at others

No Comments

No Known Activations