INDEX

Explanations

references to the word "Pitt" followed by a numeric value

oai_token-act-pair · gpt-3.5-turbo

references to the word "Pitt" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 controllers

-0.72

 Virtual

-0.70

 Aura

-0.69

 marking

-0.68

swe

-0.68

lite

-0.67

DNA

-0.66

Virtual

-0.66

 amnesty

-0.66

 erase

-0.66

POSITIVE LOGITS

 Pitt

3.85

Pitt

2.84

 Pittsburgh

1.34

 Penn

1.33

Penn

1.15

 Mann

1.13

Jol

1.12

 Panthers

1.11

Wol

1.10

 Cumber

1.05

Activations Density 0.017%

references to the word "Pitt" followed by a numeric value

references to the word "Pitt" and its variations

No Comments

No Known Activations