INDEX

Explanations

keywords related to geographical locations and historical events

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

brance

-0.76

orsi

-0.68

usercontent

-0.68

rade

-0.63

dn

-0.63

abin

-0.62

mpire

-0.61

].

-0.61

adra

-0.61

edom

-0.61

POSITIVE LOGITS

ãĥ©ãĥ³

0.62

 Moons

0.57

 boasting

0.57

BuyableInstoreAndOnline

0.56

 disbanded

0.55

 millennia

0.54

 renowned

0.54

 fierce

0.53

 sheltered

0.53

ç¥ŀ

0.52

Activations Density 0.444%

keywords related to geographical locations and historical events

No Comments

No Known Activations