apache/tvm

View on GitHub

Open Machine Learning Compiler Framework

Stars: 13.7k
Forks: 4k
Open beginner issues: 0
Indexed issues: 168
Avg merge: 2d 1h
Merged PRs (30d): 112

Dominant language: Python
License: Apache-2.0
Last GitHub push: Sep 14, 2026
Latest indexed: Sep 14, 2026
Contributing guide: No contributing guide
Code of conduct: Code of conduct
Beginner labels: No beginner labels indexed

Open All

167 open issues indexed

[Bug] [FRONTEND][ONNX] The onnx frontend incorrectly recognized two arguments for BitwiseNot operator that only has one arguement Open

needs-triage type: bug

apache/tvm#17991 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Inconcistency after meta schedule for the conv2d_hwcn operator Open

needs-triage type: bug

apache/tvm#17997 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] InstanceNorm2d -> Reciprocal -> Conv2d causing mismatch between tvm and ONNXRuntime Open

needs-triage type: bug

apache/tvm#17998 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] HardSigmoid returns 1.0 for NaN input (ONNX model mismatch with ORT) Open

needs-triage type: bug

apache/tvm#17999 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Minor output drift from ONNXRuntime in GroupNorm → Tanh → BatchNorm → Conv2d chain Open

needs-triage type: bug

apache/tvm#18000 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] When compiling using TVM's Relax, the output remains the same regardless of the optimization level set. Open

needs-triage type: bug

apache/tvm#18048 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Runtime Cross Compile Broken After The FFI Refactor Open

needs-triage type: bug

apache/tvm#18052 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Install on H100 problem about cutlass_fpA_intB_gemm Open

needs-triage type: bug

apache/tvm#18089 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] TVM result diverges from reference Open

needs-triage type: bug

apache/tvm#18100 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] InternalError: Check failed: (offset + needed_size <= this->buffer.size) is false: storage allocation failure Open

needs-triage type: bug

apache/tvm#18132 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Error occurred during tvm compilation Open

needs-triage type: bug

apache/tvm#18155 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Tracking Issue] 0.21.dev tag do not have c++ runtime inference demo Open

needs-triage type:rfc-tracking

apache/tvm#18164 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Tracking Issue] Need support for GQA Attention in Relax Open

needs-triage type:rfc-tracking

apache/tvm#18196 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [Crash][FFI/LLVM] Segfault when importing TVM after PyTorch/Transformers — LLVM static init in `COFFDirectiveParser` (`OptTable::buildPrefixChars`) Open

needs-triage type: bug

apache/tvm#18316 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[bug/feature][relax.frontend.torch] from_exported_program rejects randn.default (blocks repro that stresses advanced-indexing write + tuple output) Open

needs-triage type: bug

apache/tvm#18362 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[feature][relax.frontend.torch] Missing coverage for STFT+RNN pipeline: 'rnn_tanh.input', 'real.default', 'imag.default', 'unfold.default', 'fft_fft.default' in from_exported_program Open

needs-triage type: bug

apache/tvm#18364 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Installation process incorrect/not working Open

needs-triage type: bug

apache/tvm#18388 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [relax][torch] from_exported_program segfault with exported MHA using eq(0)/expand mask + in-place masked_fill_ (get_attr lifting warning from PyTorch) Open

needs-triage type: bug

apache/tvm#18407 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Error converting operator Attention Open

needs-triage type: bug

apache/tvm#18541 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] Performance Regression: sum operator slower on RVV than RV Open

needs-triage type: bug

apache/tvm#18560 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] Performance Issue: log operator slower on RVV Open

needs-triage type: bug

apache/tvm#18561 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] Performance Degradation: ReLU activation slower with vector extension Open

needs-triage type: bug

apache/tvm#18562 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] Performance Issue: bias_add operator slower with vectorization Open

needs-triage type: bug

apache/tvm#18563 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] sqrt operator shows poor vectorization performance Open

needs-triage type: bug

apache/tvm#18564 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] floor operator performance regression Open

needs-triage type: bug

apache/tvm#18565 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] round operator shows suboptimal vectorization Open

needs-triage type: bug

apache/tvm#18566 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] avg_pool2d operator shows performance degradation Open

needs-triage type: bug

apache/tvm#18567 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] sigmoid operator slower with vector extension Open

needs-triage type: bug

apache/tvm#18568 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] softmax operator shows suboptimal vectorization Open

needs-triage type: bug

apache/tvm#18569 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Issue: [RISC-V RVV] negative operator shows performance degradation Open

needs-triage type: bug

apache/tvm#18570 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] max_pool2d operator shows minor performance regression Open

needs-triage type: bug

apache/tvm#18571 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [RISC-V RVV] cos operator shows slight performance degradation Open

needs-triage type: bug

apache/tvm#18572 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Open

needs-triage type: bug

apache/tvm#18573 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Tracking Issue] Dogfood `tvm-ffi-stubgen` in TVM Codebase Open

needs-triage type:rfc-tracking

apache/tvm#18597 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segfault in `tvm.compile (Relax, target=llvm)` during TIR pass `InjectPTXLDG32` / `PTXRewriter::VisitStmt_(BufferStore)` even though target is CPU-only Open

needs-triage type: bug

apache/tvm#18599 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [FRONTEND][ONNX] Error converting operator ConvTranspose: InternalError: In Op(relax.add), the first input shape at dim 1 is T.int64(16) and the second input shape at dim 1 is T.int64(32), which are not broadcastable. Open

needs-triage type: bug

apache/tvm#18600 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segfault in `tvm.compile` (Relax→TIR, CUDA target) inside `tir::transform::InjectPTXLDG32` / `PTXRewriter::VisitStmt_(BufferStore)` when compiling `torch.export` model returning `(tril, triu)` tuple Open

needs-triage type: bug

apache/tvm#18612 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segfault in `tvm.compile` on **LLVM (CPU) target** when `tir.ptx_ldg32=1`: unexpectedly runs `tir::transform::InjectPTXLDG32` / `PTXRewriter` and crashes in `BufferStore` Open

needs-triage type: bug

apache/tvm#18617 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segfault in LLVM COFFOptTable when combining TVM CUDA target with PyTorch Lightning + torchmetrics in the same process Open

needs-triage type: bug

apache/tvm#18651 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segmentation fault when using TVM together with `transformers` (flan-t5-base + bfloat16 + `use_cache=True`) Open

needs-triage type: bug

apache/tvm#18653 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] Segmentation fault when importing `torchmetrics` after creating a TVM CUDA target (LLVM initialization / COFF OptTable) Open

needs-triage type: bug

apache/tvm#18655 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Release]relay ir to onnx? Open

needs-triage type:ci

apache/tvm#18679 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] LLVM version=20.0.0,compiler falied Open

needs-triage type: bug

apache/tvm#18693 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] The Onnx Frontend incorrectly mapping the Operator implementation version Open

needs-triage type: bug

apache/tvm#18698 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] compile error Open

needs-triage type: bug

apache/tvm#18708 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug][Frontend][ONNX] Split fails to handle uneven splitting with 'num_outputs' in Opset 18 Open

needs-triage type: bug

apache/tvm#18751 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[CI Problem] `Deploy / Tag tlcpackstaging to tlcpack` is failing with `unauthorized: authentication required` Open

type:ci

apache/tvm#18778 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] [Web] Batched WebGPU dispatch (#18871) causes UI jank from GPU compositor starvation Open

apache/tvm#19342 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Tracking Issue][TFlite] Expand Missing / Partial Operator Coverage Open

beginner-friendly

apache/tvm#19412 · 21 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026
[Bug] model cross compiled with cutlass can't run Open

needs-triage type: bug

apache/tvm#19422 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 14, 2026

apache/tvm

Get new issues in your inbox