INDEX

Explanations

locations and incidents related to conflict or attacks

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

 crank

-0.76

 McCarthy

-0.74

 sque

-0.72

 CEOs

-0.72

 plagiar

-0.71

 Jarrett

-0.70

 skept

-0.69

 paranormal

-0.68

 attorneys

-0.67

 smoot

-0.67

POSITIVE LOGITS

akh

1.06

inav

1.05

ulia

1.04

gam

1.00

apan

1.00

assies

0.97

opolis

0.97

ua

0.96

areth

0.96

ascus

0.94

Activations Density 0.268%

locations and incidents related to conflict or attacks

No Comments

No Known Activations