chakra
chakra copied to clipboard
mlcommons
Repository for MLCommons Chakra schema and tools
hello,I wonder if I can get an ET or some other information that can be used to generate chakra before the actual execution of the program, as I saw “We...
## Describe the Bug I am collecting inference traces using the suggested calls to PyTorch profiler and am attempting to convert them using the latest code available for Chakra. Due...
## Describe the Bug Python scripts complain about protobuf mismatch between gencode(protoc) and runtime(protobuf) versions. ## Steps to Reproduce ```bash docker run -it --rm -w /tmp ubuntu:22.04 bash #Within the...
Suppose there are 2 training jobs with their respective workload txt files. How can these be used to generate .et files so that they can be used for simulating on...
## Describe the Bug The variable ETFeederNode::num_ops_ can have uninitialized value, as it might not be initialized in the constructor. This can lead to random behavior in Workload::issue() function (see...
## Problem Related to the Feature Currently we need to manually clone and install param after running `pip install .` This causes `chakra_trace_link` to complain. Ref: https://github.com/mlcommons/chakra/wiki/Chakra-Execution-Trace-Collection-%E2%80%90-A-Comprehensive-Guide-on-Merging-PyTorch-and-Kineto-Traces#installing-chakra-and-param ## Proposed Solution...
Introduced in: https://github.com/mlcommons/chakra/pull/185 Previously, linkage was done based on the Ev idx attribute of the nodes and on the timestamps. This PR introduced linkage based on External id attribute, but...
## Describe the Bug > A clear and concise description of what the bug is. Platform:windows11 torch version:2.2.2+cu121 protobuf version:6.31.0(chakra 0.0.4 requires protobuf==5.*) it's my bug  I found that...
## Describe the Bug We followed instructions on the following wiki page: https://github.com/mlcommons/chakra/wiki/Chakra-Execution-Trace-Collection-%E2%80%90-A-Comprehensive-Guide-on-Merging-PyTorch-and-Kineto-Traces. We've captured pytorch traces and kineto traces, and the "Merging Traces with chakra_trace_link" step also went well....
