Circularity Accounting Model

CO2 increase is a main contributor to climate change, but this vast impact is not well quantified in terms of individual human behaviors. We develop tools for training CO2 sequence models for the circularity accounting model project, which develops a CO2 estimator and a CO2 production-consumption chain predictor

Our goal is to organize data so that it can be trained into models of CO2 chains.

1] Who organizes the data?

2] What is the data?

3] What are the models?

4] What are the CO2 chains?

The model architectures

The model architecture is based on front-stage image detectors and recognizers connected to production chain sequence models. this repository contains the sequence training tools.

Building our data

1. Using the carbon data from the CCaLC database and create a graph for training the network modules by using the website graph tool: https://www.entropynetwork.com/circularity/app.html?jsonURL=&xres=2048&yres=2048&imageURL=

2. Press the first line of vector, in here is ‘RawMaterials_a’ and press ‘findVector’. Copy the generated text it and paste in a new txt file.
3. Use the website to convert the generated text to the computer-can-read sentence. http://www.entropynetwork.com/text/ Paste it in the first blank box and press ‘load’. Save the generated text in the third box.

4. Do it again from the beginning using the carbon value. The final txt file should be like：

RawMaterials_a Transport1 Prodcution_a Transport2 Storage_a Transport3 Use_a RawMaterials_b Transport4 Production_b Transport5 Storage_b Transport6 Use_b Transport7

RawMaterials_a 0 Transport1 1 Prodcution_a 2 Transport2 3 Storage_a 4 Transport3 5 Use_a 6 RawMaterials_b 7 Transport4 8 Production_b 9 Transport5 10 Storage_b 11 Transport6 12 Use_b 13 Transport7 14

2105.17_a 5.89_a 597.15_a 9_a 0.00_a 9_b 620_a 2105.17_b 0.00_b 597.15_b 0.08_b 0.00_c 0.00_d 620_b 6.58_b

2105.17_a 0 5.89_a 1 597.15_a 2 9_a 3 0.00_a 4 9_b 5 620_a 6 2105.17_b 7 0.00_b 8 597.15_b 9 0.08_b 10 0.00_c 11 0.00_d 12 620_b 13 6.58_b 14

Coding section

open the terminal in your laptop. Type what inside of ‘’ ‘ssh username@xxxxxxxxxx’（connect to the server) Enter password:

1. How to put your data in the program:

’bash‘ create a new shell

’cd 2022s2s/seq2seq’ change directory to the seq2seq direc.

‘cd nmt_data’

‘cd CVM’

‘cd train’

‘ls’ (list) show all files in this folder

We can see we have four txt.file in the train folder. They correspond to the four data that we did earlier

’nano xxxx(file’s name)’ to edit the data.

Press: control + O Enter control + X Enter

Do it 4 times until all files(target.txt; sources.txt; vocab target.txt; vocab sources.txt) are filled with your data

2. How to adjust running step:

‘cd ~’

‘cd 2022s2s/seq2seq’

‘nano CVM_train.sh’

scroll down find the place where ’TRAIN_STEP = XXXXX’ Change the number to any running step you want

If your running step is not large enough, your result may appear ‘UNK’

3. How to train the model:

‘cd ~’ go back to main directory

‘cd 2022s2s/seq2seq’ re-enter the seq2seq directory

‘source ~/clone/venv3/bin/activate’ create an environment

‘./CVM_train.sh’ foreground running

‘./CVM_background_train.sh’ background running

4. How to run the model

After train the model, continue type (which means we are in the same directory and environment): ‘./CVM_1_postText.cgi’ Press enter Copy the last line of the code run it.
The result will post on the website.