INDEX

Explanations

mentions of governmental agencies or organizations named "Bureau."

oai_token-act-pair · gpt-3.5-turbo

references to various government bureaus

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

isks

-0.74

wcsstore

-0.69

vironment

-0.68

enger

-0.66

 lear

-0.65

rained

-0.65

 Hitler

-0.65

 Vader

-0.64

Putin

-0.61

DVD

-0.61

POSITIVE LOGITS

ureau

1.11

 Bureau

0.89

oreal

0.88

unal

0.73

 Headquarters

0.73

cr

0.72

xia

0.70

©¶æ¥µ

0.70

 Merrill

0.67

untu

0.67

Activations Density 0.009%

mentions of governmental agencies or organizations named "Bureau."

references to various government bureaus

No Comments

No Known Activations