INDEX

Explanations

phrases related to violent actions or dangerous items

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

 Tycoon

-0.84

 Europa

-0.70

 Horus

-0.70

 Schumer

-0.70

 charms

-0.70

Wiz

-0.69

 Scotia

-0.69

 Daniels

-0.68

 Dickinson

-0.68

 Confederation

-0.68

POSITIVE LOGITS

paren

1.18

bodied

1.13

sounding

1.06

looking

1.05

up

1.03

material

1.03

edge

1.02

average

1.02

upon

1.02

out

1.01

Activations Density 0.083%

phrases related to violent actions or dangerous items

No Comments

No Known Activations