SONG Ge

Does the usage api looks like: ```python model_dir = "hdfs://xxx.xxx.xx.xx/path/to/remote_dir" ```

> @sgwhat please take a look Sure.

Add a `transform` function to solve the error. ``` def transform(df): result = { "x": np.stack([df['f1'].to_numpy(), ..., df['f8'].to_numpy()], axis=1), "y": df['label'].to_numpy()} return result data_shard = data_shard.transform_shard(transform) est = Estimator.from_keras(keras_model=model) est.fit(data=data_shard,...

> > Add a `transform` function to solve the error. > > ``` > > def transform(df): > > result = { > > "x": np.stack([df['f1'].to_numpy(), ..., df['f8'].to_numpy()], axis=1), >...

> Create a simple but meaningful python project (e.g., multiple python source files), and use that project as a walking example in the tutorial. Sure, will commit it later.

Merged separately in #5960 , #5982 , #5985

Hey, sorry for the late reply! I have reviewed your question and the code you provided. 1. Reason. This happens because the `self.trainer.validate` was invoked in the `on_epoch_end` function. In...

> @sgwhat https://github.com/ollama/ollama/blob/main/docs/api.md#load-a-model > > This might be a good solution I have tired, but still useless.

> Thanks a lot for your works. I just tried https://github.com/dhiltgen/ollama/releases for rocm support, but I found it will be failed when using mixtral model. Here is a log for...

Hi all @airMeng @NeoZhangJianyu, I also get similar trouble with this issue. GPU device: Arc 770 System: Ubuntu ```bash sycl-ls [opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation...