INDEX

Explanations

legal terms related to laws and policies, especially those concerning penalties and abolishment

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

 Origin

-0.72

Reviewer

-0.68

 spoilers

-0.64

 Dragonbound

-0.63

 surn

-0.63

 Danish

-0.63

LCS

-0.62

 blindly

-0.62

 americ

-0.61

 deviations

-0.60

POSITIVE LOGITS

ating

1.39

uating

1.30

icating

1.29

itating

1.29

ated

1.29

ighed

1.23

uated

1.21

ited

1.21

icated

1.18

ivably

1.17

Activations Density 0.912%

legal terms related to laws and policies, especially those concerning penalties and abolishment

No Comments

No Known Activations