INDEX

Explanations

references to military, political, governmental, and structural organizations or individuals

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

ĸļ

-0.87

ĪĴ

-0.79

ulhu

-0.64

 istg

-0.59

 Lumpur

-0.59

²¾

-0.59

fif

-0.57

 subp

-0.57

wcs

-0.57

 thor

-0.56

POSITIVE LOGITS

owship

1.01

ruary

0.95

heit

0.75

oppers

0.74

ancial

0.69

illet

0.69

iencies

0.68

enhagen

0.68

antle

0.67

arlane

0.66

Activations Density 5.384%

references to military, political, governmental, and structural organizations or individuals

No Comments

No Known Activations