iceboundrock / iceboundrock/BERT4Rec
迁移基线：建立确定性 TensorFlow 1.12 oracle 与金样本
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 0
- Forks
- 0
- PR merge metrics
- No merged PRs in 30d
Description
背景
严格复现需要一个与新实现隔离的旧版行为 oracle。该 issue 不改动旧算法，只负责冻结可比较的基线。
工作项
- 建立可重建的 Python 2.7 + TensorFlow 1.12 环境，并记录镜像、CUDA/CPU 和依赖版本。
- 固定
PYTHONHASHSEED、Python random、NumPy、TensorFlow seed 以及评测候选采样 seed。 - 控制或消除
gen_data_fin.py多进程回调顺序造成的不确定性。 - 在代表性微型数据集上导出词表映射、训练/测试样本、固定候选集、模型各层输出、单步更新参数与 m/v 状态。
- 为四个现有配置记录短程训练基线及执行环境。
验收标准
- 相同环境和参数重复执行可生成相同的金样本及摘要校验值。
- 金样本足以驱动数据、模型、优化器和评测的独立对比测试。
- 基线记录包含可复现命令、seed、配置和预期指标。
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by locating gen_data_fin.py and the four existing configurations, then inspect how data generation and candidate sampling are ordered. Establish the isolated Python 2.7/TensorFlow 1.12 environment and record its versions, seeds, commands, and metrics. Done means repeated runs produce identical golden samples and checksums that support independent data, model, optimizer, and evaluation comparisons.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, tensorflow
- Domain
- machine-learning, testing
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100