INDEX

Explanations

instances of violent actions or incidents involving firearms

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Label

-0.83

Redditor

-0.76

 amen

-0.71

DonaldTrump

-0.64

arger

-0.63

 Sind

-0.63

Gi

-0.63

Else

-0.62

fty

-0.62

 Riders

-0.62

POSITIVE LOGITS

 gunshots

1.06

 bullets

1.04

 projectiles

1.04

 missiles

1.00

lessly

0.96

 volley

0.95

bomb

0.94

 arrows

0.93

 cannons

0.93

 gunfire

0.92

Activations Density 0.040%

instances of violent actions or incidents involving firearms

No Comments

No Known Activations