INDEX

Explanations

mentions of prisoners or inmates

oai_token-act-pair · gpt-3.5-turbo

mentions of "inmate" and "inmates" in the context of incarceration

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Strategy

-0.68

Sar

-0.63

 Meteor

-0.62

Dest

-0.62

Ge

-0.61

 Strat

-0.61

 Platform

-0.61

EF

-0.60

leigh

-0.60

 CONTR

-0.59

POSITIVE LOGITS

 inmates

3.49

 inmate

3.28

 prisoners

1.98

 incarcerated

1.89

 detainees

1.88

 correctional

1.78

 Correctional

1.75

 detainee

1.62

 prisons

1.58

 prisoner

1.56

Activations Density 0.014%

mentions of prisoners or inmates

mentions of "inmate" and "inmates" in the context of incarceration

No Comments

No Known Activations