INDEX

Explanations

the word "guilty" or variations of it

oai_token-act-pair · gpt-3.5-turbo

terms related to legal guilt or guilty findings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Dub

-0.74

edia

-0.69

Secure

-0.67

ffee

-0.65

oldown

-0.64

Security

-0.64

UNCH

-0.63

acco

-0.63

afety

-0.63

psey

-0.63

POSITIVE LOGITS

 verdict

1.17

 plea

1.03

 pleas

0.94

 innocence

0.87

 guilt

0.87

 conscience

0.83

 guilty

0.81

ilty

0.78

 plead

0.77

alty

0.76

Activations Density 0.040%

the word "guilty" or variations of it

terms related to legal guilt or guilty findings

No Comments

No Known Activations