tomhosking/hercules

Hercules: Attributable and Scalable Opinion Summarization (ACL 2023)

Tom Hosking

I'm a PhD student in NLP at Edinburgh supervised by Mirella Lapata, working on discrete latent variable models for language generation.

Projects

HIRO: Hierarchical Indexing for Retrieval-Augmented Opinion Summarization - Tom Hosking, Hao Tang & Mirella Lapata

Human Feedback is not Gold Standard - Tom Hosking, Phil Blunsom & Max Bartolo (ICLR 2024)

Hercules - Code for the paper "Attributable and Scalable Opinion Summarization", Tom Hosking, Hao Tang & Mirella Lapata (ACL 2023)

HRQ-VAE - Code for the paper "Hierarchical Sketch Induction for Paraphrase Generation", Tom Hosking, Hao Tang & Mirella Lapata (ACL 2022)

Separator - Code for the paper "Factorising Meaning and Form for Intent-Preserving Paraphrasing", Tom Hosking & Mirella Lapata (ACL 2021)

TorchSeq - a sequence modelling framework, built in PyTorch

McKenzie - a Slurm scheduler job monitor

My neural question generation model implementation

hercules's Issues

Dependency conflict in requirements.txt

Get the following message when trying to install requirements in new env:

`ERROR: Cannot install -r requirements.txt (line 1) and -r requirements.txt (line 2) because these package versions have conflicting dependencies.

The conflict is caused by:
allennlp 2.10.1 depends on transformers<4.21 and >=4.1
torchseq 2.4.0 depends on transformers==4.24.0`

hi test_loader not find error Q!

HI! 훈련 후 평가 과정에서 error가 발생 도움을 줄 수 있나요 ?

ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡ

교육 진행 space/amasum data
torchseq --train --reload_after_train --validate --config ./configs/hercules_space.json
교육 후 해당 폴더가 생성됨
/runs/opagg/20230816_090503_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55
평가 진행 시 error 발생
torchseq --test --load ./runs/opagg/20230816_172149_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55/
ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡ
/home/sh/anaconda3/envs/hercules/lib/python3.9/site-packages/scipy/init.py:132: UserWarning: A NumPy version >=1.21.6 and <1.28.0 is required for this version of SciPy (detected version 1.20.3)
warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"
11:25 INFO CLI ** Running with config=./runs/opagg/20230816_172149_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55/config.json **
11:25 INFO CLI ** Run ID is 20230817_112508_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55_TEST **
11:25 INFO Agent Model will run on GPU-CUDA
11:25 INFO CLI Loading from checkpoint:
11:25 INFO CLI ./runs/opagg/20230816_172149_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55/model/checkpoint.pt
11:25 INFO Agent Loading from checkpoint ./runs/opagg/20230816_172149_hrq_space_D12K12_HD005_kl25e3_init1_NL005S15_tfidf2_060ratings_hard_smarteval55/model/checkpoint.pt
11:25 INFO CLI ...loaded!
Traceback (most recent call last):
File "/home/sh/anaconda3/envs/hercules/bin/torchseq", line 8, in
sys.exit(main())
File "/home/sh/anaconda3/envs/hercules/lib/python3.9/site-packages/torchseq/main.py", line 203, in main
if data_loader.test_loader is None:
AttributeError: 'JsonDataLoader' object has no attribute 'test_loader'
ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡ
위는 error 코드입니다. 교육 진행 후 test_loader를 찾을 수 없습니다.
교육 시 생기지 않습니까?

Issue during model evaluation after setting up the virtual environment.

Recurring CUDA and GPU version compatibility issues in PyTorch. Can you tell me the CUDA version and Torch version??Recurring CUDA and GPU version compatibility issues in PyTorch. Can you tell me the CUDA version and Torch version??

Issues regarding pip installation and model evaluation

Hi,
I am trying to rerun Hercules from my server but there are two issues:

1. There is a warning message WARNING: Did not find branch or tag 'a6ebc10', assuming revision or ref.. Not sure whether it might lead to some potential issues.
2. If we ignore that warning message, the installation seems ok. However, I get an error message 'JsonDataLoader' object has no attribute 'test_loader' while running torchseq --test --load ./models/hercules_space/

Any thoughts regarding this? Thanks very much.

AmaSum dataset download issue

I can't download the AmaSum dataset. Is there a way to obtain it?

Download links broken

Download links for Space broken. Can't run eval without opagg/space-eval/test.jsonl which is not available through repo data folder (space-eval subfolder is not in opagg folder) and can't download through broken link in README under "Download Data/Models"