CUDA Device-Side Assert from Tokenizer/Embedding Vocab Mismatch
indexSelectLargeIndex: srcIndex < srcSelectDimSize assertion fires when input token ids exceed the embedding table size. The standard outcome of pairing a tokenizer that has added tokens with a model whose embeddings were never resized.
indexSelectLargeIndex: srcIndex < srcSelectDimSize assertion fires when input token ids exceed the embedding table size. The standard outcome of pairing a tokenizer that has added tokens with a model whose embeddings were never resized.
What this failure is
CUDA Device-Side Assert from Tokenizer/Embedding Vocab Mismatch is a Training Stability failure seen during ML training runs. indexSelectLargeIndex: srcIndex < srcSelectDimSize assertion fires when input token ids exceed the embedding table size. The standard outcome of pairing a tokenizer that has added tokens with a model whose embeddings were never resized. Common tags: Device Assert, Cuda, Tokenizer, Embedding.
Is this what broke your run? Paste your log.
You're reading about CUDA Device-Side Assert from Tokenizer/Embedding Vocab Mismatch. 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)
Len(tokenizer) > model.config.vocab_size after add_special_tokens/add_tokens. Loading a different run's tokenizer with overlapping name. Label ids beyond num_labels for classification heads (same assert family). 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
- CUDA error: device-side assert triggered, unusable context afterward
- assert text names Indexing.cu / indexSelectLargeIndex
- reproduces on specific data shards (the ones containing new tokens)
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| Assertion srcIndex < srcSelectDimSize failed | len(tokenizer) > model.config.vocab_size after add_special_tokens/add_tokens |
| crash moves with the dataset, not the GPU | loading a different run's tokenizer with overlapping name |
| CPU run of the same batch raises IndexError: index out of range | label ids beyond num_labels for classification heads (same assert family) |
Which systems are affected
- HF transformers fine-tunes with added special tokens
- PEFT/LoRA flows that swap tokenizers
- multi-run pipelines sharing tokenizer artifacts
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: Assertion srcIndex < srcSelectDimSize failed
- ✓Verified signal present: crash moves with the dataset, not the GPU
- ✓Verified signal present: CPU run of the same batch raises IndexError: index out of range
- ✓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
- len(tokenizer) > model.config.vocab_size after add_special_tokens/add_tokens
- loading a different run's tokenizer with overlapping name
- label ids beyond num_labels for classification heads (same assert family)
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.