Is this a hardware or a software issue?

Synchronization failures can look like either layer depending on the symptom. The discriminator is the resolution of the symptom to the recommended fix: if applying the steps under "How to fix it" eliminates the symptom, this is a software/configuration class issue regardless of the hardware it surfaced on. Repeat the fix on a fresh node to confirm.

How long does the fix typically take?

Most resolutions from this entry take 10 to 30 minutes of engineer time once the diagnosis is confirmed. Long fixes come from skipping the diagnosis step and chasing apparent symptoms (validation drift, memory growth, slow throughput) instead of the actual root cause listed above. Apply the fix in a small experiment first, then roll it forward.

Does this affect inference or only training?

Primarily this is a synchronization failure observed during training, but the same misconfiguration can leak into downstream evaluation and serving. Treat the prevention checklist as a guardrail: any pipeline that touches a trained checkpoint should re-validate the failure mode before promoting the model.

What's the cheapest way to confirm this is the problem?

Cheapest signal: reproduce the symptom on a 1-GPU toy run with the same configuration. If the toy run exhibits the same behavior, the cause is the configuration rather than the cluster. If only the multi-GPU run is affected, the failure is a distributed interaction, so PyTorch is the most likely layer to inspect.

Can the Denpex platform detect this automatically?

Yes. Denpex correlates the symptoms in this entry against your training logs in real time and surfaces the failure taxonomy, root cause, and exact remediation command on the console. Three full diagnoses are included free during signup, and the encyclopedia entry you are reading is generated from the same signal library that powers the platform.

What related failures should I check next?

Closely related failures cluster around #Autograd Graph Deadlock. After applying this fix, validate against the related entries below to confirm no overlapping cause is masking as this one. Engineers routinely discover that the dominant failure mode is actually a second, sibling failure compounded on top.

Will this recur after the fix?

Applied once with the prevention checklist, this failure does not recur under normal operation. The recurring variant most engineers encounter is the re-emergence of the same symptom on a different model or stack after Autograd Graph Deadlock is upgraded. Pin compatible versions and rerun the prevention smoke test on every dependency change.

Is there a fingerprint I can monitor?

Yes. The training script completely freezes during the backward pass. is the highest-signal fingerprint. Wire a metric or alert on that signal (or its absence after the fix) so you can detect the recurrence automatically. The Denpex platform encodes this exact fingerprint and surfaces it directly in the dashboard.

What is the official documentation that covers this?

Reference documentation is listed in the References section at the bottom of this entry. The most authoritative URLs are the upstream framework docs and the issue trackers cited; the Denpex entry synthesizes both into the executive, engineer, and on-call perspectives so each audience can act without reading every linked source.

Does this entry include working code?

Yes. The "Code Examples" section contains copy-paste-ready snippets that target the most common stacks (PyTorch, Hugging Face Trainer, DeepSpeed, Megatron-LM, vLLM/Transformers). Each snippet is annotated with why it works so you can adapt the pattern to your own framework without losing fidelity.

DDP Hang Due to Unused Parameters in Forward Pass

Q: What does this failure look like in production?

In production this failure typically appears during the first tens-to-hundreds of training steps: loss curves become non-monotonic, validation metrics plateau above the baseline, and downstream checkpoints either diverge or fail to improve. Engineers frequently misread it as data drift when the underlying mechanism is when find_unused_parameters=false (the default), ddp assumes every parameter in the model will receive a gradient. if a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. ddp waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step.. Validating against the symptom list in this entry is the fastest way to confirm the diagnosis.

Q: How does this scale from 1 GPU to 1000 GPUs?

At single-GPU scale the symptom usually looks like slow convergence or a single anomalous loss value. At 64+ GPU scale, distributed interactions amplify when find_unused_parameters=false (the default), ddp assumes every parameter in the model will receive a gradient. if a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. ddp waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step. through collectives, and the failure can manifest as a full job abort. Always reproduce the fix at the smallest cluster configuration that still exhibits the symptom before promoting to production scale.

When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. DDP waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step.

Quick answer

When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient.

Symptom: RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one.
Root cause: When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g.
Recommended fix: Enable find_unused_parameters if the graph is truly dynamic. model = DDP(model, device_ids=[rank], find_unused_parameters=True) Instructs DDP to traverse the autograd graph before the backward pass to discover which parameters are unused and proactively mark them as ready for reduction.
How Denpex helps: Denpex investigates DDP Hang Due to Unused Parameters in Forward Pass using the evidence you provide or your connected workload collects. Earlier rank, host or application evidence is needed to distinguish an initiating failure from a downstream report.

Synchronization#Autograd Graph Deadlock

What this failure is

DDP Hang Due to Unused Parameters in Forward Pass is a Synchronization failure seen during ML training runs. When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. DDP waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step. Common tags: Autograd Graph Deadlock.

Live diagnosis, no signup

Is this what broke your run? Paste your log.

You're reading about DDP Hang Due to Unused Parameters in Forward Pass. Paste your own traceback and relevant evidence for an investigation of your workload, with a next action or a specific missing fact. A reference entry does not establish your cause. No account or card for the free diagnosis. Review data handling before submitting sensitive logs.

Before uploading, review cloud data handling and local options.

training_logs.txt

No log to hand? Try one:

Upload or drop a plain-text log

Upload your log file. Large files are processed in your browser. Only selected evidence is submitted when you run diagnosis.

3 free diagnoses/day

Why it happens (the mechanism)

The script produces no crash or error message immediately. It simply hangs. Naive diagnosis might assume an NCCL network issue, GPU hardware failure, or a slow data loader, whereas the true issue is entirely in the autograd computation graph.

What you'll observe

RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one.
Process hangs indefinitely during loss.backward()

Common symptoms and what they mean

Symptom	Why it happens
The training script completely freezes during the backward pass.	When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. DDP waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step.
No CPU or GPU utilization spikes; processes just sit idle waiting for communication.	When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. DDP waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step.

Which systems are affected

PyTorch
DistributedDataParallel (DDP)

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.

✓Set environment variable TORCH_DISTRIBUTED_DEBUG=DETAIL to see exact parameter wait logs.
✓Check if any layers in the model's forward pass are inside conditional blocks that might not execute.
✓Temporarily set find_unused_parameters=True to see if the hang resolves.

Root cause

When find_unused_parameters=False (the default), DDP assumes every parameter in the model will receive a gradient. If a parameter is skipped in the forward pass (e.g., due to a conditional if-statement), its gradient is never computed. DDP waits indefinitely in the backward pass for the all-reduce operation on that missing gradient, deadlocking the training step.

The fix and how to prevent it

Searchable error signature

search key

RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one.

Use this text as a lookup key in logs and upstream issue trackers. It is not presented as a captured customer log. Confirm the cause from your own preceding events, versions, configuration and the cited references.

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.

No 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 extension

References

↗PyTorch Distributed Troubleshooting: Unused Parameters

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.

Diagnose your logs free Compare a solved incident Browse all failures

Related Synchronization errors

DDP Hang from Uneven Dataset Sizes Across Ranks

Synchronization · critical

SyncBatchNorm Deadlock During Single-Rank Evaluation

Synchronization · high

dist.all_gather Timeout from Dynamic Tensor Shapes

Synchronization · high