INDEX

Explanations

percentages and numerical increases

oai_token-act-pair · gpt-3.5-turbo

statistics and numerical increases in data

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

 incompet

-0.86

 Poor

-0.82

 Shame

-0.80

 disgust

-0.77

ueless

-0.77

 obscurity

-0.76

 crappy

-0.76

Poor

-0.75

 misunderstand

-0.73

 shitty

-0.72

POSITIVE LOGITS

 increase

1.38

 surge

1.29

 doubling

1.21

 uptick

1.21

 increases

1.20

 tripled

1.16

 Increase

1.15

 growth

1.14

 Increases

1.14

 quadru

1.12

Activations Density 0.521%

percentages and numerical increases

statistics and numerical increases in data

No Comments

No Known Activations