INDEX

Explanations

words related to security classifications or sensitive information

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

amus

-0.66

·

-0.62

absolute

-0.62

shown

-0.61

 Twain

-0.60

bane

-0.60

anon

-0.60

 Couch

-0.60

\\\\\\\\

-0.59

Ķ

-0.58

POSITIVE LOGITS

ised

1.25

ties

1.21

izations

1.11

isations

1.09

ized

0.99

ty

0.92

ises

0.91

ities

0.91

izes

0.89

 relativity

0.89

Activations Density 0.065%

words related to security classifications or sensitive information

No Comments

No Known Activations