INDEX

Explanations

personal pronouns and verb forms referring to actions by or involving people

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

 Millennium

-0.74

 Consortium

-0.68

 Globe

-0.66

 Eleven

-0.65

Eps

-0.64

 Vers

-0.63

 Aston

-0.60

 Around

-0.60

 Morgan

-0.59

 Greatest

-0.58

POSITIVE LOGITS

've

1.68

'd

1.54

're

1.51

'll

1.47

'm

1.19

 knew

1.19

 encount

1.15

 could

1.11

 intend

1.08

can

1.08

Activations Density 3.195%

personal pronouns and verb forms referring to actions by or involving people

No Comments

No Known Activations