Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update GPeft example Readme #812

Merged
merged 4 commits into from
Apr 24, 2024
Merged

Update GPeft example Readme #812

Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
6671e49
Fix GPeft readme
Apr 24, 2024
1a83bbb
Update
Apr 24, 2024
40c327b
Update
Apr 24, 2024
876f25c
Merge branch 'main' into fix-gpeft-llm-readme
classicsong Apr 24, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 17 additions & 5 deletions examples/peft_llm_gnn/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,16 @@
## Preparing the environment
Please follow https://graphstorm.readthedocs.io/en/latest/install/env-setup.html to setup your GraphStorm environment.
In addition, run the following scripts to install necessary python packages

```
pip install ipython
pip install peft
```

## Preparing Amazon Review dataset
This folder contains the data processing script to process the raw Amazon Review dataset
downloaded from https://cseweb.ucsd.edu/~jmcauley/datasets/amazon_v2/. We use domain Video
Games https://datarepo.eng.ucsd.edu/mcauley_group/data/amazon_v2/metaFiles2/meta_Video_Games.json.gz
Games https://datarepo.eng.ucsd.edu/mcauley_group/data/amazon_v2/metaFiles2/meta_Video_Games.json.gz
and put it under raw_data as an example.
```
wget https://datarepo.eng.ucsd.edu/mcauley_group/data/amazon_v2/metaFiles2/meta_Video_Games.json.gz \
Expand All @@ -15,21 +25,21 @@ python preprocess_amazon_review.py

Once the data are processed, run the following command to construct a graph
for PEFT LLM-GNNs in GraphStorm for node classification on level-3 product type.
The command takes `AR_Video_Games.json` that specifies the input data for graph
The command takes `AR_Video_Games.json` that specifies the input data for graph
construction, constructs the graph, and saves the parition to `amazon_review`.

```
python -m graphstorm.gconstruct.construct_graph \
--conf-file AR_Video_Games.json \
--output-dir datasets/amazon_review_Video_Games/ \
--graph-name amazon_review \
--num-processes 16 --num-parts 1 \
--num-processes 16 --num-parts 1 \
--skip-nonexist-edges --add-reverse-edges

```

## Train LLM-GNN model to predict product type of items
The command below runs parameter-efficient fine-tuning of LLM-GNNs on node
The command below runs parameter-efficient fine-tuning of LLM-GNNs on node
classification and link prediction via `main_nc.py` and `main_lp.py`.


Expand All @@ -38,6 +48,8 @@ classification and link prediction via `main_nc.py` and `main_lp.py`.
WORKSPACE=$PWD
dataset=amazon_review
domain=Video_Games
cp -r datasets/amazon_review_"$domain" datasets/amazon_review_nc_"$domain"

python3 -m graphstorm.run.launch \
--workspace "$WORKSPACE" \
--part-config datasets/amazon_review_nc_"$domain"/amazon_review.json \
Expand All @@ -59,7 +71,7 @@ dataset=amazon_review
domain=Video_Games
python -m graphstorm.run.launch \
--workspace "$WORKSPACE" \
--part-config "$WORKSPACE"/dataset/amazon_review_"$domain"/amazon_review.json \
--part-config "$WORKSPACE"/datasets/amazon_review_"$domain"/amazon_review.json \
--ip-config ./ip_list.txt \
--num-trainers 8 \
--num-servers 1 \
Expand Down
Loading