wb14123/seq2seq-couplet

在训练的时候出现了这个错误，Found Inf or NaN global norm，哪个大牛帮忙看一下怎么破？

Open

#25 opened on Jan 24, 2019

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Python (5,477 stars) (1,072 forks)batch import
help wantedquestion

Description

InvalidArgumentError (see above for traceback): Found Inf or NaN global norm. : Tensor had Inf values [[node VerifyFinite/CheckNumerics (defined at C:\Users\xxx\seq2seq-couplet-master1\model.py:79) = CheckNumericsT=DT_FLOAT, message="Found Inf or NaN global norm.", _device="/job:localhost/replica:0/task:0/device:GPU:0"]]

Contributor guide

Tech stack
python
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.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
basic TensorFlow knowledgeunderstanding of gradient issues
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
The issue reports 'Found Inf or NaN global norm' during training of a seq2seq couplet model. The error originates at model.py line 79 in the global norm calculation. Investigate potential causes such as unstable gradients, learning rate too high, or issues in the loss function. Consider adding gradient clipping or adjusting optimizer parameters. The only comment is from the author asking for help; no maintainer response has been provided. Check the repository's training scripts for hyperparameters and data preprocessing steps that might lead to numerical instability.