INDEX

Explanations

URL shortening services and fraudulent content keywords, particularly related to bribes

oai_token-act-pair · gpt-3.5-turbo

mentions of the website "usat" and terms related to bribery and carrots

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

 Gallery

-0.77

eries

-0.68

lake

-0.66

tem

-0.66

 Disco

-0.65

 Sanct

-0.64

 Mast

-0.63

syn

-0.63

sections

-0.63

POSITIVE LOGITS

 bribe

2.44

 bribes

2.43

 carrot

2.05

usat

1.67

 carrots

1.63

 bribery

1.61

 deterrent

1.39

 coercion

1.38

 persuasion

1.35

 incentive

1.27

Activations Density 0.044%

URL shortening services and fraudulent content keywords, particularly related to bribes

mentions of the website "usat" and terms related to bribery and carrots

No Comments

No Known Activations