INDEX

Explanations

mentions of Twitter usernames or handles, particularly focusing on "JP."

oai_token-act-pair · gpt-3.5-turbo

references to the entity "JP" or related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

rights

-0.85

leneck

-0.73

 Hurricanes

-0.72

 Rebels

-0.68

 thirds

-0.67

 Semin

-0.67

querque

-0.66

conf

-0.65

 Masquerade

-0.63

 Musk

-0.63

POSITIVE LOGITS

JP

1.65

EG

1.03

KR

0.96

NK

0.94

JP

0.93

PLIC

0.90

IME

0.87

EF

0.87

NJ

0.85

oint

0.85

Activations Density 0.007%

mentions of Twitter usernames or handles, particularly focusing on "JP."

references to the entity "JP" or related terms

No Comments

No Known Activations