INDEX

Explanations

terms related to office spaces and activities

oai_token-act-pair · gpt-3.5-turbo

mentions of "office."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oise

-0.71

rous

-0.67

rd

-0.65

âķĲâķĲ

-0.62

ï¸

-0.62

 Adin

-0.61

icult

-0.61

Beg

-0.60

Â©

-0.60

 QUEST

-0.59

POSITIVE LOGITS

office

1.02

holders

1.01

holder

1.00

 office

0.93

ĳå£«

0.83

person

0.82

matical

0.81

©¶æ

0.79

 offices

0.79

endar

0.78

Activations Density 0.022%

terms related to office spaces and activities

mentions of "office."

No Comments

No Known Activations