INDEX

Explanations

terms related to the word "cripple."

oai_token-act-pair · gpt-3.5-turbo

references to a specific location, "Cripple," or related terms

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

 masse

-0.80

 reperto

-0.72

ercise

-0.68

eri

-0.67

um

-0.66

 Kare

-0.65

 Metall

-0.64

 Antar

-0.63

 exerc

-0.61

izen

-0.61

POSITIVE LOGITS

ripp

3.40

ursed

1.60

ooks

1.12

embed

1.11

KO

1.04

 Flat

0.98

 Hollow

0.98

uds

0.94

ilipp

0.91

urd

0.89

Activations Density 0.048%

terms related to the word "cripple."

references to a specific location, "Cripple," or related terms

No Comments

No Known Activations