INDEX

Explanations

phrases related to decision-making or judgment

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

 guessing

-0.67

 Thoughts

-0.63

 Cards

-0.60

 Anyway

-0.60

 boarding

-0.59

 Doctrine

-0.58

 Gallery

-0.57

 COUR

-0.56

ONE

-0.56

hat

-0.56

POSITIVE LOGITS

 accompanies

1.17

 arose

1.08

 preceded

1.01

 encompasses

0.98

 enables

0.98

 resulted

0.97

 resembles

0.97

 promotes

0.95

 governs

0.94

 occurs

0.94

Activations Density 1.562%

phrases related to decision-making or judgment

No Comments

No Known Activations