INDEX

Explanations

phrases related to law enforcement and criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to individuals who have committed crimes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

dust

-0.72

sonian

-0.71

quickShipAvailable

-0.71

erd

-0.70

pher

-0.69

printed

-0.69

 Pupp

-0.68

conn

-0.68

ports

-0.66

porting

-0.66

POSITIVE LOGITS

 offender

0.84

 ineligible

0.84

 offenders

0.73

 sentenced

0.72

 adjud

0.71

 sanction

0.70

 sentencing

0.68

 abusing

0.67

alties

0.66

 mitigation

0.66

Activations Density 0.039%

phrases related to law enforcement and criminal activities

references to individuals who have committed crimes

No Comments

No Known Activations