INDEX

Explanations

references to a specific person or character named "Chan" with varying relevance based on the activation value

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific individual, likely a prominent figure

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

sylv

-0.76

ilit

-0.65

 geological

-0.65

solid

-0.64

 interstate

-0.63

forces

-0.62

ob

-0.62

 Militia

-0.61

 unlawfully

-0.60

bleacher

-0.60

POSITIVE LOGITS

 Chan

4.07

Chan

2.81

 Chung

1.47

 Chong

1.45

 Chow

1.37

 Wong

1.35

 Chun

1.28

 Cham

1.25

chan

1.25

Lau

1.25

Activations Density 0.022%

references to a specific person or character named "Chan" with varying relevance based on the activation value

mentions of a specific individual, likely a prominent figure

No Comments

No Known Activations