[Bug] [Module Name] Bug title 如何提高知识抽取时的并发数，现在始终保持在5个并发，速度太慢了。

[su400]

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Windows

What happened

1、如何提高知识抽取时的并发数，现在始终保持在5个并发，速度太慢了，请告知在哪里设置。
2、构建完知识库后，构建时所采用的模型就不能更换，在全局设置里更换无效。我预期的场景是使用商业模型构建图谱，推理时采用本地模型，希望能够更换。

How to reproduce

v.0.6版本，win11

Are you willing to submit PR?

• Yes I am willing to submit a PR!

[zzyyll2]

docker-compose.yml里有个配置，你改一下看是否有效：

[caszkgui]

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Windows

What happened

1、如何提高知识抽取时的并发数，现在始终保持在5个并发，速度太慢了，请告知在哪里设置。 2、构建完知识库后，构建时所采用的模型就不能更换，在全局设置里更换无效。我预期的场景是使用商业模型构建图谱，推理时采用本地模型，希望能够更换。

How to reproduce

v.0.6版本，win11

Are you willing to submit PR?

• Yes I am willing to submit a PR!

In developer mode, User can set concurrency by modifying num_chain and num_threads_per_chain in kag_config.yaml:

If you are dealing with one large size documents，you can increase the num of num_threads_per_chain. In other scenarios such as hotpotqa dataset, which stores multi items in one json file, you can increase the num of num_chain.

[caszkgui]

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Windows

What happened

1、如何提高知识抽取时的并发数，现在始终保持在5个并发，速度太慢了，请告知在哪里设置。 2、构建完知识库后，构建时所采用的模型就不能更换，在全局设置里更换无效。我预期的场景是使用商业模型构建图谱，推理时采用本地模型，希望能够更换。

How to reproduce

v.0.6版本，win11

Are you willing to submit PR?

• Yes I am willing to submit a PR!

In developer mode, user can set multi LLM model service in different stage.
For example, in example_config.yaml, KAG provide the openie_llm and chat_llm, which are referenced in kag-builder-pipeline and kag-solver-pipeline separately.

[someone132s]

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Windows

What happened

1、如何提高知识抽取时的并发数，现在始终保持在5个并发，速度太慢了，请告知在哪里设置。 2、构建完知识库后，构建时所采用的模型就不能更换，在全局设置里更换无效。我预期的场景是使用商业模型构建图谱，推理时采用本地模型，希望能够更换。

How to reproduce

v.0.6版本，win11

Are you willing to submit PR?

• Yes I am willing to submit a PR!

In developer mode, user can set multi LLM model service in different stage. For example, in example_config.yaml, KAG provide the openie_llm and chat_llm, which are referenced in kag-builder-pipeline and kag-solver-pipeline separately.

您好，我在linux下使用docker部署，我想知道这个配置文件在哪呢？