xxyp

你好，请问怎么样使用deepspeed多卡训练呀，加了deepspeed config后跑不起来

7

│ 771 │ │ if self.distributed_type == DistributedType.DEEPSPEED: │ │ ❱ 772 │ │ │ config = self.deepspeed_plugin.deepspeed_config │ │ 773 │ │ │ if config.get("fp16", {}).get("enabled", False): │ │...

本地rocksdb，执行stop-hugegraph.sh后，再执行start-hugegraph.sh，为什么数据会丢失呢

1

### Bug Type (问题类型) None ### Before submit - [X] 我已经确认现有的 [Issues](https://github.com/apache/hugegraph/issues) 与 [FAQ](https://hugegraph.apache.org/docs/guides/faq/) 中没有相同 / 重复问题 (I have confirmed and searched that there are no similar problems in the...

使用docker部署hugegraph，挂载了宿主机目录，本地RocksDB ，版本0.11.0，使用一段时间后，写数据提示no space left on device，清理了之后还是不行

1

### Bug Type (问题类型) None ### Before submit - [X] 我已经确认现有的 [Issues](https://github.com/apache/hugegraph/issues) 与 [FAQ](https://hugegraph.apache.org/docs/guides/faq/) 中没有相同 / 重复问题 (I have confirmed and searched that there are no similar problems in the...

训练后的模型infer的时候报输入形状错误 RuntimeError: Tensors must have same number of dimensions: got 4 and 2

4

训练后的模型infer的时候报输入形状错误 outputs = self.generate(**input_ids, **gen_kwargs) RuntimeError: Tensors must have same number of dimensions: got 4 and 2

容器里可以pull代码仓，通过前端分析的时候，拉取代码失败 main server 报错：'cd_error': [ErrorDetail(string='代码库及帐号不匹配', code='invalid')] scmproxy.log报错： unable to access 代码仓 The requested URL returned error: 504

speed up inference?

1

employing Codestral-22B-v0.1 by Mistral-inference, the speed seems slow. there any methods to speed up the inference ?