tvm
tvm copied to clipboard
apache
Open deep learning compiler stack for cpu, gpu and specialized accelerators
### Expected behavior TVM should build the model correctly. ### Actual behavior ```c Traceback (most recent call last): File "/home/carla/Documents/test_tvm/0321/test_relax2.py", line 75, in tvm_model = relax.transform.LegalizeOps()(tvm_model) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/carla/Documents/tvm/python/tvm/ir/transform.py", line...
Fixed bug where CUDA codegen produces faulty code when a vectorizable BufferLoadNode contains a Float8 type. Codegen generated the invalid signature `make___nv_fp8x2_e5m2(param_0[v_.x], param_0[v_.y])` where "param_0" is of type `__nv_fp8_e5m2* __restrict__`....
### Expected behavior TVM should build the model correctly. ### Actual behavior ```c Traceback (most recent call last): File "/home/carla/Documents/test_tvm/0312/test_relax2.py", line 81, in ex = relax.build(tvm_model, target="llvm") ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/carla/Documents/tvm/python/tvm/relax/vm_build.py",...
### Expected behavior TVM should build the model correctly. ### Actual behavior ```c Traceback (most recent call last): File "/home/carla/Documents/test_tvm/0312/test_relax2.py", line 81, in ex = relax.build(tvm_model, target="llvm") ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/carla/Documents/tvm/python/tvm/relax/vm_build.py",...
### Expected behavior The onnx frontend should import the model correctly. ### Actual behavior ```c Error converting operator Transpose, with inputs: [R.where(v0_0, v3_0, lv14)] Traceback (most recent call last): File...
### Expected behavior Build tvm-0.18.0 from source. ### Actual behavior Hi, I follow this document (https://tvm.apache.org/docs/install/from_source.html) to install tvm 0.18.0 from source. Everything was fine until I ran the command:...
### Expected behavior I expect each axis will be used only one time when it generate sketch design ### Actual behavior I try to print the sketch design in function...
I had previously raised this issue on MLC LLM as well, but it seems that the root cause lies in `PagedKVCache`. With the recent release of Phi-4-mini-inst, the introduction of...
### Overview This PR supports warp-level shuffle primitives using the newly introduced `subgroup` in WebGPU. We then use them in the implementation of allreduce lowering. The introduced primitives are: -...
The image name has a typo and there is no image tagged with latest on the docker hub. This should be fixed and is likely an issue with all the...
