INDEX

Explanations

instances of the word "law" and related terms in the context of law enforcement

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

 Tate

-0.68

 Cinnamon

-0.64

 Cros

-0.63

hetic

-0.63

Dak

-0.60

ACTED

-0.60

 Rubber

-0.59

 Wast

-0.59

 Gree

-0.59

xit

-0.58

POSITIVE LOGITS

 enforcement

1.53

suit

1.37

enforcement

1.31

suits

1.28

 Enforcement

1.27

lessness

1.27

yer

1.24

breakers

1.19

yers

1.16

 abiding

1.15

Activations Density 0.326%

instances of the word "law" and related terms in the context of law enforcement

No Comments

No Known Activations