INDEX

Explanations

negative or harmful words and phrases

oai_token-act-pair · gpt-3.5-turbo

negative descriptors related to harm or damage

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ħ¢

-0.80

£ı

-0.72

orthy

-0.69

ŃĶ

-0.67

uana

-0.66

ebus

-0.65

å§«

-0.63

rity

-0.63

ajo

-0.62

 reinstated

-0.61

POSITIVE LOGITS

lessly

0.75

ishly

0.71

 inhib

0.70

ingly

0.69

uminati

0.69

 wasteful

0.67

umin

0.66

monkey

0.65

 relegation

0.65

 negativity

0.65

Activations Density 0.652%

negative or harmful words and phrases

negative descriptors related to harm or damage

No Comments

No Known Activations