INDEX

Explanations

terms related to legal actions, specifically arrest warrants

oai_token-act-pair · gpt-3.5-turbo

references to legal warrants

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

angles

-0.65

 races

-0.63

lins

-0.62

elight

-0.62

Sl

-0.61

opolis

-0.61

isu

-0.59

<+

-0.58

effects

-0.58

iang

-0.58

POSITIVE LOGITS

 warrant

4.10

 warrants

3.33

 Warrant

2.51

 warranted

1.57

 subpoena

1.32

 WARRANT

1.31

 merit

1.27

 justify

1.15

 permit

1.14

 affidavit

1.13

Activations Density 0.021%

terms related to legal actions, specifically arrest warrants

references to legal warrants

No Comments

No Known Activations