INDEX

Explanations

mentions of prisoners, particularly in a war or conflict setting

oai_token-act-pair · gpt-3.5-turbo

references to prisoners, particularly in the context of war or captivity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

drive

-0.72

lag

-0.71

OPA

-0.71

 Boll

-0.68

orie

-0.67

wig

-0.67

 Ples

-0.66

 Quadro

-0.65

ernaut

-0.64

 Beir

-0.64

POSITIVE LOGITS

 prisoners

0.92

 incarcerated

0.82

 inmates

0.82

 detainees

0.82

 prisoner

0.82

 sentenced

0.82

 captives

0.77

zees

0.72

 freed

0.72

icts

0.70

Activations Density 0.047%

mentions of prisoners, particularly in a war or conflict setting

references to prisoners, particularly in the context of war or captivity

No Comments

No Known Activations