INDEX

Explanations

phrases related to legal terms or court proceedings

oai_token-act-pair · gpt-3.5-turbo

references to legal convictions

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

ks

-0.72

iti

-0.67

pher

-0.65

atered

-0.64

bies

-0.63

parts

-0.63

sites

-0.63

 Skies

-0.60

 Interactive

-0.60

DR

-0.60

POSITIVE LOGITS

 conviction

3.86

 convictions

3.00

 apprehension

1.54

 convict

1.53

 indictment

1.49

 belief

1.47

 convicted

1.45

 prosecution

1.38

 prosecutions

1.35

conv

1.33

Activations Density 0.019%

phrases related to legal terms or court proceedings

references to legal convictions

No Comments

No Known Activations