INDEX

Explanations

mentions of criminal activity and legal proceedings

oai_token-act-pair · gpt-3.5-turbo

discussions surrounding legal cases involving vulnerable individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 honored

-1.04

 favors

-0.97

 folks

-0.94

 labor

-0.92

 honoring

-0.91

 flavor

-0.90

 modernization

-0.89

 favorably

-0.89

avored

-0.88

 favor

-0.88

POSITIVE LOGITS

 Â£

1.51

 paed

1.47

 offences

1.38

 CCTV

1.33

 counselling

1.29

 neighbour

1.26

NHS

1.25

 solicitor

1.25

 maths

1.24

 Cardiff

1.23

Activations Density 0.625%

mentions of criminal activity and legal proceedings

discussions surrounding legal cases involving vulnerable individuals

No Comments

No Known Activations