INDEX

Explanations

names of people or places, potentially when associated with specific qualities or titles

oai_token-act-pair · gpt-3.5-turbo

proper nouns and brand names

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

Not in Any Lists

No Comments

Negative Logits

 charact

-0.71

NetMessage

-0.68

 abnormal

-0.67

entry

-0.66

perty

-0.61

orously

-0.61

 safer

-0.60

 damp

-0.60

 condition

-0.60

scribed

-0.58

POSITIVE LOGITS

 Productions

1.03

 himself

0.94

 herself

0.88

 Enterprises

0.87

 Consulting

0.86

 Presents

0.83

 Himself

0.83

 Films

0.81

 Yourself

0.80

 Associates

0.79

Activations Density 0.310%

names of people or places, potentially when associated with specific qualities or titles

proper nouns and brand names

No Comments

No Known Activations