INDEX

Explanations

words relating to principles, ethics, or moral considerations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

 myster

-0.78

 vulner

-0.76

 sacrific

-0.74

 limb

-0.74

 mathemat

-0.73

 writ

-0.73

 trainers

-0.70

 conduc

-0.69

 builders

-0.69

 destro

-0.68

POSITIVE LOGITS

ï¸ı

1.31

vernment

1.04

SpaceEngineers

0.95

lean

0.95

log

0.92

ove

0.91

ï¸

0.91

ËĪ

0.90

deg

0.89

âĹ¼

0.89

Activations Density 0.036%

words relating to principles, ethics, or moral considerations

No Comments

No Known Activations