INDEX

Explanations

references to political figures addressed as "Minister"

oai_token-act-pair · gpt-3.5-turbo

references to prime ministers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

crew

-0.65

Osc

-0.63

MpServer

-0.63

aukee

-0.61

 battlefield

-0.61

 directors

-0.61

 subpoena

-0.58

artisan

-0.57

 lethal

-0.56

 Anchorage

-0.56

POSITIVE LOGITS

 Narendra

1.24

 Shinzo

1.21

 Justin

0.96

 Benjamin

0.93

ial

0.92

 Theresa

0.91

Prime

0.90

 Recep

0.87

 Stephen

0.87

 Tony

0.85

Activations Density 0.040%

references to political figures addressed as "Minister"

references to prime ministers

No Comments

No Known Activations