INDEX

Explanations

phrases related to fear and potential negative consequences

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aepernick

-0.81

Favorite

-0.80

idelines

-0.75

nice

-0.73

ichick

-0.73

Cola

-0.71

isms

-0.70

issance

-0.67

MQ

-0.67

brates

-0.66

POSITIVE LOGITS

 jeopard

1.31

 repercussions

1.28

 repr

1.21

 retribution

1.17

 impending

1.16

 damaging

1.12

 consequences

1.09

 future

1.07

 harm

1.06

 endanger

1.05

Activations Density 2.001%

phrases related to fear and potential negative consequences

No Comments

No Known Activations