INDEX

Explanations

phrases indicating intent or purpose related to criminal or harmful actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

visor

-0.78

 Chocobo

-0.72

visors

-0.69

 Cham

-0.67

 Jenner

-0.67

 Serge

-0.66

Jub

-0.66

 Britann

-0.65

 Tycoon

-0.65

 Jackets

-0.64

POSITIVE LOGITS

ful

0.94

ensity

0.89

ality

0.86

fulness

0.86

 intent

0.86

uality

0.82

edly

0.80

ually

0.80

eous

0.79

ruction

0.78

Activations Density 0.009%

phrases indicating intent or purpose related to criminal or harmful actions

No Comments

No Known Activations