dbiir/UER-py

When using sentencepiece, a Segmentation fault is reported

Open

#278 opened on Mar 1, 2022

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Python (520 forks)github user discovery
help wanted

Repository metrics

Stars
 (3,108 stars)
PR merge metrics
 (No merged PRs in 30d)

Description

Word-based pretraining with sentencepiece

python3 preprocess.py --corpus_path corpora/book_review.txt \
                      --spm_model_path models/cluecorpussmall_spm.model \
                      --dataset_path book_review_word_sentencepiece_dataset.pt \
                      --processes_num 8 --seq_length 128 --dynamic_masking \
                      --data_processor mlm

python3 pretrain.py --dataset_path book_review_word_sentencepiece_dataset.pt \
                    --spm_model_path models/cluecorpussmall_spm.model \
                    --output_model_path models/book_review_word_sentencepiece_model.bin \
                    --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
                    --total_steps 5000 --save_checkpoint_steps 2500 --report_steps 500 \
                    --learning_rate 1e-4 --batch_size 64 \
                    --tie_weights

Report the following error Segmentation fault

Contributor guide

Research direction
Reproduce the segmentation fault with the provided commands. Use 'ulimit c unlimited' and GDB to get a backtrace. Check sentencepiece library version compatibility and memory usage (e.g., batch size, seq length). Try reducing batch size or processes num to isolate the issue. Verify that the spm model file is valid and loaded correctly.
Tech stack
pythonpytorch
Domain
machine learning
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
PythonPyTorch
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
60