Direct LLM Calls

Don’t wait for reasoning on basic jobs.

What is Direct LLM Call As the capacity of LLMs increases, their ability to handle more complex requests in a single call also grows significantly. LLMs, once limited to making only a few tool calls, can now perform hundreds of operations and engage in deep thinking. In the BeaconLabs framework, we approach this scenario as follows:

• Agents are LLMs that are specifically designed and focused on a particular domain.

• Agents should be used when there's a need to understand the organization’s objectives, goals, and rules.

However, beyond these two cases, many other tasks can be handled directly by an LLM. This is where the Direct LLM Call feature comes in.

Making a Direct LLM Call To send a task directly to an LLM without any intermediary layers, you can use the Direct object. It works with the same interface as agents, making it easy to switch or design based on your needs.

pythonCopyEditfrom beaconlabs import Task, Direct

task = Task("Create History paper of LLMs")

Direct.print_do(task)
# or Direct.do(task)  # This one