INDEX

Explanations

negation and contradictions

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

ãĤ¼

-0.78

kamp

-0.71

éĥ

-0.71

çļ

-0.68

ously

-0.67

udeau

-0.67

 Tours

-0.64

 Calls

-0.64

USH

-0.64

çīĪ

-0.64

POSITIVE LOGITS

 necessarily

1.25

 gonna

1.23

hin

1.21

icable

1.18

epad

1.11

icably

1.07

eworthy

1.04

 really

0.99

 bothered

0.99

 bothering

0.99

Activations Density 0.086%

negation and contradictions

No Comments

No Known Activations