INDEX

Explanations

terms related to the Justice Department

oai_token-act-pair · gpt-3.5-turbo

references to the Justice Department

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

 stub

-0.72

 altitude

-0.69

insk

-0.66

 isol

-0.66

 bumps

-0.64

 tunnel

-0.64

 starve

-0.63

 whisk

-0.62

atars

-0.62

 simulate

-0.62

POSITIVE LOGITS

 Justice

3.99

Justice

3.34

 JUSTICE

2.55

justice

2.50

 justice

2.48

DOJ

1.89

 justices

1.72

Judge

1.57

 Attorney

1.53

Attorney

1.43

Activations Density 0.016%

terms related to the Justice Department

references to the Justice Department

No Comments

No Known Activations