{"record":{"id":"460dd0d378116128","repo":"tracel-ai/burn","slug":"todo-quantization-not-supported-yet-460dd0","errorCode":null,"errorMessage":"todo!(\"Quantization not supported yet\")","messagePattern":"todo!\\(\"Quantization not supported yet\"\\)","errorType":"panic","errorClass":null,"httpStatus":null,"severity":"error","filePath":"crates/burn-fusion/src/ops/transaction.rs","lineNumber":21,"sourceCode":"    ops::{TransactionOps, TransactionPrimitive},\n};\n\nuse crate::{Fusion, FusionBackend};\n\nimpl<B: FusionBackend> TransactionOps<Fusion<B>> for Fusion<B> {\n    async fn tr_execute(\n        transaction: TransactionPrimitive<Self>,\n    ) -> Result<burn_backend::ops::TransactionPrimitiveData, ExecutionError> {\n        B::tr_execute(TransactionPrimitive::new(\n            transaction\n                .read_floats\n                .into_iter()\n                .map(|t| t.client.clone().resolve_tensor_float::<B>(t))\n                .collect(),\n            transaction\n                .read_qfloats\n                .into_iter()\n                .map(|_t| todo!(\"Quantization not supported yet\"))\n                .collect(),\n            transaction\n                .read_ints\n                .into_iter()\n                .map(|t| t.client.clone().resolve_tensor_int::<B>(t))\n                .collect(),\n            transaction\n                .read_bools\n                .into_iter()\n                .map(|t| t.client.clone().resolve_tensor_bool::<B>(t))\n                .collect(),\n        ))\n        .await\n    }\n}\n","sourceCodeStart":3,"sourceCodeEnd":37,"githubUrl":"https://github.com/tracel-ai/burn/blob/d16f7ba2ed0d41408189384044cc886fb4c8f957/crates/burn-fusion/src/ops/transaction.rs#L3-L37","documentation":"The fusion backend's tr_execute (crates/burn-fusion/src/ops/transaction.rs:21) resolves each read tensor to its client representation, but for read_qfloats it maps every element to todo!(\"Quantization not supported yet\"). The fusion server has no quantized-tensor handling, so any fused transaction containing a quantized read panics.","triggerScenarios":"Executing a transaction (tr_execute) on the fusion backend where transaction.read_qfloats contains at least one QuantizedTensor handle.","commonSituations":"Running a model with quantized weights/layers under the fusion (JIT) backend and batching reads via a transaction; quantization works through normal ops but not the fused transaction path.","solutions":["Do not include quantized tensors in fusion transactions; convert them to float tensors (dequantize) before building the transaction.","Perform quantized reads individually, outside of TransactionPrimitive.","Check the fusion backend's feature matrix for quantization support before adopting quantized ops."],"exampleFix":null,"handlingStrategy":"validation","validationCode":"assert!(transaction.read_qfloats.is_empty(), \"fusion backend does not support quantized reads in transactions\");","typeGuard":null,"tryCatchPattern":null,"preventionTips":["Keep quantized tensors out of fused transactions on the fusion backend.","Prefer float representations for batched reads under fusion.","Track upstream fusion quantization support before adopting quantized layers."],"tags":["rust","burn","fusion","quantization","unimplemented"],"backgroundTag":"quantization-unsupported","analyzedSha":"d16f7ba2ed0d41408189384044cc886fb4c8f957","analyzedAt":"2026-09-05T13:19:14.260Z","contentChangedAt":"2026-09-05T13:19:14.260Z","schemaVersion":2},"datasetVersion":"2026-09-12T17:17:11.597Z"}