INDEX

Explanations

references to the TV show "Black Mirror"

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Mirror."

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

 Atkins

-0.88

 odor

-0.79

buf

-0.69

keye

-0.67

asse

-0.65

 conv

-0.63

ieg

-0.63

atical

-0.63

throp

-0.62

 legalized

-0.62

POSITIVE LOGITS

 Mirror

4.06

 mirror

2.03

Mir

1.91

mir

1.61

 mirrors

1.58

 mirrored

1.22

 Herald

1.11

Mir

1.11

 Mail

1.11

 Telegraph

1.09

Activations Density 0.014%

references to the TV show "Black Mirror"

mentions of the word "Mirror."

No Comments

No Known Activations