Generative models to create network configurations through natural language prompts
In one implementation, a device obtains a natural language-based description of a network via a user interface. The device generates, based on the natural language-based description, network configuration parameters for the network using a generative model. The device conducts a simulation of traffic in the network using the network configuration parameters, to obtain telemetry data. The device uses the telemetry data to train a machine learning model to perform network analytics.
1 . A method comprising:
obtaining, by a device, a natural language-based description of a network via a user interface, the natural language-based description comprising one or more user specified aspects of the network;
inputting, by the device, the natural language-based description into a generative model trained to create network configurations from natural language-based descriptions; and
generating, by the device and using the generative model based on the natural language-based description, network configuration parameters for the network, wherein the generative model generates realistic values for one or more non-specified aspects from the natural language-based description of the network in order to generate the network configuration parameters.
2 . The method as in claim 1 , wherein the generative model comprises a large language model.
3 . The method as in claim 1 , wherein the natural language-based description of the network indicates a geographic location of the network, and wherein the network configuration parameters are based in part on the geographic location.
4 . The method as in claim 1 , wherein generating the network configuration parameters comprises:
receiving user feedback from the user interface regarding the network configuration parameters; and
using reinforcement learning to refine the network configuration parameters based on the user feedback.
5 . The method as in claim 1 , wherein the network configuration parameters comprise a latency or loss distribution for the network.
6 . The method as in claim 1 , wherein generating the network configuration parameters comprises:
mapping previously collected telemetry data from one or more other networks to the network configuration parameters and wherein the generative model is trained to associate the network configuration parameters for the one or more other networks to natural language-based descriptions for the one or more other networks.
7 . The method as in claim 1 , further comprising:
conducting, by the device, a simulation of traffic in the network using the network configuration parameters, to obtain telemetry data; and
using, by the device, the telemetry data to train a machine learning model to perform network analytics.
8 . The method as in claim 7 , further comprising:
deploying, by the device, the machine learning model for execution by a network controller or networking device.
9 . The method as in claim 7 , wherein the network analytics comprise at least one of:
performing network anomaly detection, performing network what-if analysis, or performing network troubleshooting, or performing network predictions.
10 . The method as in claim 1 , wherein the natural language-based description of the network indicates a topology of the network.
11 . An apparatus, comprising:
one or more network interfaces;
a processor coupled to the one or more network interfaces and configured to execute one or more processes; and
a memory configured to store a process that is executable by the processor, the process when executed configured to:
obtain a natural language-based description of a network via a user interface, the natural language-based description comprising one or more user specified aspects of the network;
input the natural language-based description into a generative model trained to create network configurations from natural language-based descriptions; and
generate, using the generative model and based on the natural language-based description, network configuration parameters for the network, wherein the using a generative model generates realistic values for one or more non-specified aspects from the natural language-based description of the network in order to generate the network configuration parameters.
12 . The apparatus as in claim 11 , wherein the generative model comprises a large language model.
13 . The apparatus as in claim 11 , wherein the natural language-based description of the network indicates a geographic location of the network, and wherein the network configuration parameters are based in part on the geographic location.
14 . The apparatus as in claim 11 , wherein the apparatus generates the network configuration parameters by:
receiving user feedback from the user interface regarding the network configuration parameters; and
using reinforcement learning to refine the network configuration parameters based on the user feedback.
15 . The apparatus as in claim 11 , wherein the network configuration parameters comprise a latency or loss distribution for the network.
16 . The apparatus as in claim 11 , wherein the apparatus generates the network configuration parameters by:
mapping previously collected telemetry data from one or more other networks to the network configuration parameters and wherein the generative model is trained to associate the network configuration parameters for the one or more other networks to natural language-based descriptions for the one or more other networks.
17 . The apparatus as in claim 11 , further comprising:
conducting a simulation of traffic in the network using the network configuration parameters, to obtain telemetry data; and
using the telemetry data to train a machine learning model to perform network analytics.
18 . The apparatus as in claim 17 , wherein the process when executed is further configured to:
deploy the machine learning model for execution by a network controller or networking device.
19 . The apparatus as in claim 17 , wherein the network analytics comprise at least one of:
performing network anomaly detection, performing network what-if analysis, or performing network troubleshooting, or performing network predictions.
20 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:
obtaining, by the device, a natural language-based description of a network via a user interface, the natural language-based description comprising one or more user specified aspects of the network;
inputting, by the device, the natural language-based description into a generative model trained to create network configurations from natural language-based descriptions; and
generating, by the device and using the generative model based on the natural language-based description, network configuration parameters for the network, wherein the generative model generates realistic values for one or more non-specified aspects from the natural language-based description of the network in order to generate the network configuration parameters.