INDEX

Explanations

references to inmates or prison-related terms

oai_token-act-pair · gpt-3.5-turbo

mentions of inmates and corrections-related terms

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

IFrame

Link

Negative Logits

rous

-0.73

ãĤ¤ãĥĪ

-0.71

ãĥ¼ãĥ³

-0.70

ku

-0.68

drive

-0.67

ãĥ¼ãĥ«

-0.67

 Nare

-0.67

orically

-0.66

efully

-0.66

issan

-0.66

POSITIVE LOGITS

 inmates

0.93

 inmate

0.92

icts

0.91

iaries

0.90

iary

0.85

 Facility

0.75

 incarcerated

0.72

 Correctional

0.72

 correctional

0.70

arians

0.70

Activations Density 0.026%

references to inmates or prison-related terms

mentions of inmates and corrections-related terms

No Known Activations