INDEX

Explanations

negative or harmful actions and characteristics

oai_token-act-pair · gpt-3.5-turbo

negative emotions and destructive 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-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

•GPT2 Layer 8 Movie Sentiment Features

No Comments

Negative Logits

Ħ¢

-0.81

£ı

-0.81

ŃĶ

-0.76

ebus

-0.74

orthy

-0.70

uana

-0.68

glas

-0.68

orney

-0.64

Grab

-0.63

ĩ

-0.62

POSITIVE LOGITS

uminati

0.79

ingly

0.79

lessly

0.75

ãĥ¼

0.74

monkey

0.73

ãĥĨ

0.71

 ridden

0.69

ishly

0.69

 miser

0.68

fully

0.67

Activations Density 0.618%

negative or harmful actions and characteristics

negative emotions and destructive actions

No Comments

No Known Activations