Keqiu Hu

Results 16 issues of Keqiu Hu

# Describe your issue - Is this a bug report or a feature request? bug report - Describe the issue tap device doesn't respond - What is the expected behaviour?...

Current TF binding is not compatible with TF 2.4+, reference: #401. This PR follows recent changes in [Horovod](https://github.com/horovod/horovod/blob/7b44897733e724ba2a077e3bb277bd481e5be006/horovod/_keras/__init__.py#L117) to update distributed optimizer code to be compatible.

**Describe the bug** The example `tensorflow2_keras_mnist.py` example is not compatible with TF 2.5. **To Reproduce** Steps to reproduce the behavior: 1. install tensorflow-gpu (2.5) 2. install byteps latest (0.2.5) ```...

(Creating this issue for visibility so people interested can join the discussion... ) ## Overview Load Apache ORC formatted data natively into TensorFlow from file system supported by TensorFlow, e.g....

We want to have the retry id in the container environment to help the recover process. Initial run: [] First retry: [RETRY_ID:0] Second retry: [RETRY_ID:1]

good first issue

``` 19/08/31 22:35:28 ERROR tony.ApplicationMaster: Received error in AM to RM call java.lang.NullPointerException: Task was null! Nothing to schedule. at com.google.common.base.Preconditions.checkNotNull(Preconditions.java:900) at com.linkedin.tony.ApplicationMaster$ContainerLauncher.run(ApplicationMaster.java:1061) at com.linkedin.tony.ApplicationMaster$RMCallbackHandler.onContainersAllocated(ApplicationMaster.java:1019) at org.apache.hadoop.yarn.client.api.async.impl.AMRMClientAsyncImpl$CallbackHandlerThread.run(AMRMClientAsyncImpl.java:366) ``` in code...

add byteps environment set up to TonY

We rely on testTonyE2E for almost all TonyClient tests, that introduces a lot of overhead. As we are setting all necessary context in TonyClient, we should make it easier to...

# Prerequisites Please answer the following questions for yourself before submitting an issue. - [x] I am running the latest code. Development is very rapid so there are no tagged...

performance
AMD GPU

As part of https://github.com/numba/llvmlite/pull/976 discussion, split the `llvmdev` build into a separate PR for easier testing. Validated on Ubuntu, it builds `llvm` with `compiler-rt` properly. ``` ❯ tar xvf llvmdev-14.0.6-h84b9e52_3.tar.bz2...

4 - Waiting on reviewer