INDEX

Explanations

phrases related to criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to legal or criminal situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Supported

-0.68

oday

-0.65

SPONSORED

-0.64

MpServer

-0.62

iscons

-0.62

WAS

-0.62

BT

-0.62

arij

-0.61

HUD

-0.61

Access

-0.60

POSITIVE LOGITS

 nowadays

1.00

 dictators

0.89

 presidents

0.86

 rarely

0.85

 usually

0.81

 sometimes

0.79

 tends

0.79

 seldom

0.79

 endings

0.77

 Usually

0.76

Activations Density 0.894%

phrases related to criminal activities

references to legal or criminal situations

No Comments

No Known Activations