INDEX

Explanations

phrases related to crime and legal proceedings

oai_token-act-pair · gpt-3.5-turbo

references to crime and violence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 yourselves

-0.62

speak

-0.58

 denotes

-0.56

 recognizes

-0.54

iple

-0.53

 DOES

-0.53

 Definitions

-0.53

!:

-0.51

 concise

-0.50

wise

-0.50

POSITIVE LOGITS

 allegedly

0.60

 nightclub

0.60

 Olympia

0.59

 assailant

0.56

 injuring

0.55

 assailants

0.54

 illegally

0.54

 ecstasy

0.53

stasy

0.53

 overrun

0.52

Activations Density 1.880%

phrases related to crime and legal proceedings

references to crime and violence

No Comments

No Known Activations