INDEX

Explanations

names or terms related to various entities or individuals

oai_token-act-pair · gpt-3.5-turbo

the mentions of names or terms related to an individual with the name "ame."

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

İĭ

-0.83

keyes

-0.80

 stocking

-0.74

ĻĤ

-0.70

 mathemat

-0.69

 stabilized

-0.68

 slender

-0.68

thening

-0.67

ŃĶ

-0.66

 necks

-0.65

POSITIVE LOGITS

ame

1.57

ames

1.11

AME

0.87

leon

0.86

ilial

0.85

emn

0.82

amed

0.82

 Shame

0.77

Ã±a

0.75

aming

0.73

Activations Density 0.005%

names or terms related to various entities or individuals

the mentions of names or terms related to an individual with the name "ame."

No Comments

No Known Activations