INDEX

Explanations

phrases related to negative outcomes or disappointments

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

rouse

-0.76

utenberg

-0.71

zag

-0.70

igate

-0.69

 kindred

-0.68

afort

-0.67

arthy

-0.64

area

-0.64

avia

-0.63

glas

-0.63

POSITIVE LOGITS

 alas

0.79

 sadly

0.78

 disappoint

0.74

 unfortunately

0.74

 untrue

0.70

 forgot

0.70

 misunderstand

0.70

terday

0.68

 unfortunate

0.68

nown

0.67

Activations Density 0.034%

phrases related to negative outcomes or disappointments

No Comments

No Known Activations