INDEX

Explanations

words related to social and political issues, critical comments, and mentions of crises

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 VIDEOS

-0.71

 Doodle

-0.68

 subsequent

-0.65

 Flavoring

-0.63

 preceding

-0.60

ppings

-0.60

regate

-0.57

 accompanies

-0.57

 sake

-0.57

 subsequently

-0.56

POSITIVE LOGITS

ivated

0.90

fed

0.88

ivating

0.87

fortable

0.87

enced

0.87

alysed

0.85

itialized

0.84

 ready

0.83

cerned

0.81

lined

0.81

Activations Density 6.003%

words related to social and political issues, critical comments, and mentions of crises

No Comments

No Known Activations