Add G-retriever (GNN+LLM) example

[puririshi98]

1. Integrate LLM class #9462
2. Add nn.models.GRetriever #9480
3. Add WebQSPDataset #9481
4. -> Add G-retriever (GNN+LLM) example #9167

---

repro:
Latest NVIDIA PyG container
+
git config --global credential.helper store; huggingface-cli login; cd /opt/pyg; pip uninstall -y torch-geometric; rm -rf pytorch_geometric; git clone -b gnn-llm-model-integration https://github.com/pyg-team/pytorch_geometric.git; cd /opt/pyg/pytorch_geometric; pip install .; pip install peft datasets transformers pcst_fast sentencepiece; python3 examples/llm_plus_gnn/g_retriever.py

old PR: #9154

note: pure cpu is 220x slower than pure GPU using a single Grace Hopper (for llama-7b)

info:
tried gemma, performs worse in all train/val/test metrics. most likely needs some tuning, will leave this as future work as part of the community sprint to try many LLM and GNN combos and tune them. Therefore keeping the default llama2

the new gemma-v2 is also much worse than llama2

[puririshi98]

@Kh4L reviews welcome

[codecov]

Codecov Report

Attention: Patch coverage is 16.10738% with 250 lines in your changes are missing coverage. Please review.

Project coverage is 87.21%. Comparing base (61c47ee) to head (48be260).
Report is 10 commits behind head on master.

❗ Current head 48be260 differs from pull request most recent head 6e0151f

Please upload reports for the commit 6e0151f to get more accurate results.

Files	Patch %	Lines
torch_geometric/nn/text/llm.py	16.41%	112 Missing ⚠️
torch_geometric/nn/models/g_retriever.py	11.66%	106 Missing ⚠️
...eometric/nn/text/sentence_transformer_embedding.py	20.00%	32 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #9167      +/-   ##
==========================================
- Coverage   87.33%   87.21%   -0.13%     
==========================================
  Files         460      477      +17     
  Lines       30385    31051     +666     
==========================================
+ Hits        26536    27080     +544     
- Misses       3849     3971     +122     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[puririshi98]

@akihironitta @rusty1s addressed reviews, plz lmk if anything else needed to merge.

[akihironitta]

Sorry for the extreme delay :( Will have a look during this weekend (long weekend in UK) :)

[puririshi98]

@akihironitta anything else needed to merge?

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Please upload report for BASE (web-qsp-integ@87bd353). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
torch_geometric/nn/nlp/llm.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@               Coverage Diff                @@
##             web-qsp-integ    #9167   +/-   ##
================================================
  Coverage                 ?   87.31%           
================================================
  Files                    ?      481           
  Lines                    ?    31345           
  Branches                 ?        0           
================================================
  Hits                     ?    27369           
  Misses                   ?     3976           
  Partials                 ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.