Import / Environment Error
Import and environment errors crash training at startup due to missing or mismatched dependencies.
Import and environment errors crash training at startup due to missing or mismatched dependencies.
What this failure is
Import / Environment Error is a Environment failure seen during ML training runs. Import and environment errors crash training at startup due to missing or mismatched dependencies. Common tags: Import Error, Environment, Dependencies, Cuda.
Is this what broke your run? Paste your log.
You're reading about Import / Environment Error. Paste your own crash log or traceback below and get the real root cause for YOUR run, not this generic entry. No account, no card. Logs are masked at ingress and never saved to account history.
Want 14 days on the Scale plan?
Request an evaluation code. A verified workplace organization activates up to 50 diagnoses a day, alerts, history, and follow-up questions. No credit card or automatic subscription.
Why it happens (the mechanism)
PyTorch-CUDA version mismatch: PyTorch built for CUDA 12.1 but system has CUDA 11.8 drivers installed. Container image inconsistency across nodes: one node has an older cuDNN version or missing NCCL library. Python path pollution: a previous activation or module load added incompatible paths to sys.path. Shared library conflicts: LD_LIBRARY_PATH includes directories with incompatible versions of libcudart, libnccl, or libcublas. Taken together, these mechanisms explain why the failure is reproducible, why it tends to surface on specific workloads or scales, and why generic mitigation attempts often fall short without addressing the underlying cause.
What you'll observe
- Training script fails at import time
- Different ranks succeed inconsistently
- Previously working training fails after update
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| ModuleNotFoundError: No module named 'torch' | PyTorch-CUDA version mismatch: PyTorch built for CUDA 12.1 but system has CUDA 11.8 drivers installed |
| ImportError: libcudart.so.12: cannot open shared object file | Container image inconsistency across nodes: one node has an older cuDNN version or missing NCCL library |
| RuntimeError: Detected that PyTorch and CUDA are not compatible | Python path pollution: a previous activation or module load added incompatible paths to sys.path |
| Segfault at Python interpreter startup with no Python traceback | Shared library conflicts: LD_LIBRARY_PATH includes directories with incompatible versions of libcudart, libnccl, or libcublas |
Which systems are affected
- SLURM jobs with environment modules or containerized training
- Kubernetes pods with custom Docker images
- Python venv / conda environments with conflicting packages
- Multi-node training where nodes have different software stacks
How to confirm this is the problem
Use this checklist to test the hypothesis against a small reproduction. No single line proves the root cause, so preserve the preceding events and compare one variable at a time.
- ✓Reproduce the failure from a clean checkpoint/seed: the symptom must appear without warm-up state from a previous run.
- ✓Verified signal present: ModuleNotFoundError: No module named 'torch'
- ✓Verified signal present: ImportError: libcudart.so.12: cannot open shared object file
- ✓Verified signal present: RuntimeError: Detected that PyTorch and CUDA are not compatible
- ✓Verified signal present: Segfault at Python interpreter startup with no Python traceback
- ✓A targeted fix from the "How to fix it" section eliminates or substantially reduces the symptom within one validation pass.
The fix and the prevention pattern
The root cause is on this page and stays free. A free account adds the exact remediation steps, saved history, and the fix on every entry in the encyclopedia.
Sign up free. Unlock the full analysisNo credit card. Daily allowance follows verified trust tier. Instant access.
Diagnose this failure in VS Code
Select the traceback or open the failed terminal, then run Denpex locally to see the initiating rank, collateral failures, exact fix, and verification command without uploading the log.
Install the free VS Code extensionCUDA errors in context
CUDA reports errors asynchronously, so the traceback usually points at whatever line synchronised next rather than the one at fault. The hub covers every common CUDA error and how to make it report honestly.
Compare every cuda error side by sideRelated failures to investigate next
Root cause
- PyTorch-CUDA version mismatch: PyTorch built for CUDA 12.1 but system has CUDA 11.8 drivers installed
- Container image inconsistency across nodes: one node has an older cuDNN version or missing NCCL library
- Python path pollution: a previous activation or module load added incompatible paths to sys.path
- Shared library conflicts: LD_LIBRARY_PATH includes directories with incompatible versions of libcudart, libnccl, or libcublas
The fix and how to prevent it
Evaluate Denpex on your own logs
Request a Scale evaluation code. A verified workplace organization activates 14 days with up to 50 diagnoses a day. Every account keeps its current diagnosis allowance and gets a verification path. No card or automatic subscription.
Don't just read the fix, diagnose your run
The encyclopedia tells you what went wrong. Denpex tells you what went wrong in YOUR training run. With your logs, your config, and your stack.