Revert "[Unity] Implement relax.transform.KillAfterLastUse (#15810)" #15852
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
)" This reverts commit aa4587f. Unfortunately, this PR broke MLC LLM's build pipeline, more specifically, the command below ```python python3 -m mlc_llm.build --model dist/models/llama-2-13b-chat-hf/ --quantization q4f16_1 ``` leads to the the following error message: <details> ``` Traceback (most recent call last): File "<frozen runpy>", line 198, in _run_module_as_main File "<frozen runpy>", line 88, in _run_code File "/opt/scratch/junrushao/mlc-llm/mlc_llm/build.py", line 13, in <module> main() File "/opt/scratch/junrushao/mlc-llm/mlc_llm/build.py", line 10, in main core.build_model_from_args(parsed_args) File "/opt/scratch/junrushao/mlc-llm/mlc_llm/core.py", line 616, in build_model_from_args new_params = utils.convert_weights(param_manager, params, args) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/mlc-llm/mlc_llm/utils.py", line 258, in convert_weights vm["transform_params"]() File "/opt/scratch/junrushao/tvm-dev/python/tvm/_ffi/_ctypes/packed_func.py", line 239, in __call__ raise_last_ffi_error() File "/opt/scratch/junrushao/tvm-dev/python/tvm/_ffi/base.py", line 476, in raise_last_ffi_error raise py_err File "/opt/scratch/junrushao/tvm-dev/src/runtime/relax_vm/vm.cc", line 634, in tvm::runtime::relax_vm::VirtualMachineImpl::InvokeClosurePacked(tvm::runtime::ObjectRef const&, tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) clo->impl.CallPacked(TVMArgs(values.data(), tcodes.data(), args.size() + 1), rv); ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/src/runtime/relax_vm/vm.cc", line 708, in operator() *rv = static_cast<VirtualMachineImpl*>(ctx_ptr)->InvokeBytecode(gf_idx, inputs); ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/src/runtime/relax_vm/vm.cc", line 765, in tvm::runtime::relax_vm::VirtualMachineImpl::InvokeBytecode(long, std::vector<tvm::runtime::TVMRetValue, std::allocator<tvm::runtime::TVMRetValue> > const&) RunLoop(); File "/opt/scratch/junrushao/tvm-dev/src/runtime/relax_vm/vm.cc", line 890, in tvm::runtime::relax_vm::VirtualMachineImpl::RunLoop() this->RunInstrCall(curr_frame, instr); ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/src/runtime/relax_vm/vm.cc", line 843, in tvm::runtime::relax_vm::VirtualMachineImpl::RunInstrCall(tvm::runtime::relax_vm::VMFrame*, tvm::runtime::relax_vm::Instruction) this->InvokeClosurePacked(func_pool_[instr.func_idx], args, &ret); ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/mlc-llm/mlc_llm/relax_model/param_manager.py", line 579, in set_item loaded_params[i] = tvm.nd.array(computed_param, device=device_cpu) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/python/tvm/runtime/ndarray.py", line 635, in array return empty(arr.shape, arr.dtype, device, mem_scope).copyfrom(arr) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/python/tvm/runtime/ndarray.py", line 390, in empty dtype = DataType(dtype) ^^^^^^^^^^^^^^^ File "/opt/scratch/junrushao/tvm-dev/python/tvm/_ffi/runtime_ctypes.py", line 174, in __init__ raise ValueError("Do not know how to handle type %s" % type_str) ValueError: Do not know how to handle type object ``` </details> To briefly explain the root cause of this issue, `set_item` method, as defined [here](https://github.com/mlc-ai/mlc-llm/blob/4f4a93f03fed3d900605c02de575d7d5f429ed79/mlc_llm/relax_model/param_manager.py#L576-L579), gets `computed_param=None` after this commit. As a temporary solution, I'd love to propose that we revert this commit for now to quickly unblock us from building any LLM model, but I'm happy to get it back immediately as soon as the issue is fixed.
Member
Author
Contributor
|
Thank you, and I've identified why this happens. The Working on two fixes now, one for the short-term and one for the long-term. The short-term fix will add similar handling to |
Contributor
|
The short-term PR is submitted at #15854, which adds another special-case in |
Member
Author
|
Thanks for quickly acting on this! Let’s get #15854 merged instead! |
Lunderberg
added a commit
to Lunderberg/tvm
that referenced
this pull request
Oct 2, 2023
Prior to this commit, if a relax variable were assigned to itself, through either `VarBinding` or `MatchCast` nodes, the two relax variables would share the same register in the VM. As a result, any upstream transform that deletes an object with `R.memory.kill_tensor` or `R.memory.kill_storage` must be aware of this VM behavior, and to only output one such instruction for each set of aliased registers. This commit updates the VM to produce one register for each aliased relax variable. The trivial bindings can be removed applying `relax.transform.CanonicalizeBindings`, instead of being implicitly de-duplicated at the codegen level. This PR is a follow-up to apache#15854, with a better long-term solution, but which may have knock-on effects that must also be resolved. In addition, this adds a usage example for the bug reported in apache#15852, to avoid re-occurrence of similar issues.
Contributor
|
Sounds good! I now also have #15855 prepared, which applied the more general fix and can be merged in at a more leisurely pace. |
Lunderberg
added a commit
to Lunderberg/tvm
that referenced
this pull request
Oct 3, 2023
Prior to this commit, if a relax variable were assigned to itself, through either `VarBinding` or `MatchCast` nodes, the two relax variables would share the same register in the VM. As a result, any upstream transform that deletes an object with `R.memory.kill_tensor` or `R.memory.kill_storage` must be aware of this VM behavior, and to only output one such instruction for each set of aliased registers. This commit updates the VM to produce one register for each aliased relax variable. The trivial bindings can be removed applying `relax.transform.CanonicalizeBindings`, instead of being implicitly de-duplicated at the codegen level. This PR is a follow-up to apache#15854, with a better long-term solution, but which may have knock-on effects that must also be resolved. In addition, this adds a usage example for the bug reported in apache#15852, to avoid re-occurrence of similar issues.
Lunderberg
added a commit
to Lunderberg/tvm
that referenced
this pull request
Oct 24, 2023
Prior to this commit, if a relax variable were assigned to itself, through either `VarBinding` or `MatchCast` nodes, the two relax variables would share the same register in the VM. As a result, any upstream transform that deletes an object with `R.memory.kill_tensor` or `R.memory.kill_storage` must be aware of this VM behavior, and to only output one such instruction for each set of aliased registers. This commit updates the VM to produce one register for each aliased relax variable. The trivial bindings can be removed applying `relax.transform.CanonicalizeBindings`, instead of being implicitly de-duplicated at the codegen level. This PR is a follow-up to apache#15854, with a better long-term solution, but which may have knock-on effects that must also be resolved. In addition, this adds a usage example for the bug reported in apache#15852, to avoid re-occurrence of similar issues.
Lunderberg
added a commit
to Lunderberg/tvm
that referenced
this pull request
Oct 25, 2023
Prior to this commit, if a relax variable were assigned to itself, through either `VarBinding` or `MatchCast` nodes, the two relax variables would share the same register in the VM. As a result, any upstream transform that deletes an object with `R.memory.kill_tensor` or `R.memory.kill_storage` must be aware of this VM behavior, and to only output one such instruction for each set of aliased registers. This commit updates the VM to produce one register for each aliased relax variable. The trivial bindings can be removed applying `relax.transform.CanonicalizeBindings`, instead of being implicitly de-duplicated at the codegen level. This PR is a follow-up to apache#15854, with a better long-term solution, but which may have knock-on effects that must also be resolved. In addition, this adds a usage example for the bug reported in apache#15852, to avoid re-occurrence of similar issues.
Lunderberg
added a commit
that referenced
this pull request
Oct 31, 2023
* [Unity] Ensure one VM register for each relax binding Prior to this commit, if a relax variable were assigned to itself, through either `VarBinding` or `MatchCast` nodes, the two relax variables would share the same register in the VM. As a result, any upstream transform that deletes an object with `R.memory.kill_tensor` or `R.memory.kill_storage` must be aware of this VM behavior, and to only output one such instruction for each set of aliased registers. This commit updates the VM to produce one register for each aliased relax variable. The trivial bindings can be removed applying `relax.transform.CanonicalizeBindings`, instead of being implicitly de-duplicated at the codegen level. This PR is a follow-up to #15854, with a better long-term solution, but which may have knock-on effects that must also be resolved. In addition, this adds a usage example for the bug reported in #15852, to avoid re-occurrence of similar issues. * Add unit test to validate that the alias is preserved * Use the new GetBoundValue utility function * Update VMTIRCodeGen to also avoid de-duplication of bindings * Move the callback definition to tvm.relax.testing.vm namespace
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This reverts commit aa4587f.
Unfortunately, this PR broke MLC LLM's build pipeline, more specifically, the command below
leads to the the following error message:
Details
To briefly explain the root cause of this issue,
set_itemmethod, as defined here, getscomputed_param=Noneafter this commit.As a temporary solution, I'd love to propose that we revert this commit for now to quickly unblock us from building any LLM model, but I'm happy to get it back immediately as soon as the issue is fixed.