v0.2.1

feat: Optimize historical friend nodes connected at startup
fix: Forgot to clean up the expired data in the peer_collect_db when refactoring the timer
feat: One node supports multiple models with the same name, which are distinguished by container id of docker
feat: Add registration/unregistration interface for a single container model
fix: Model error may lead to dead loop forwarding
docs: Update documentations

v0.2.0

feat: The get interface uses query data to parse parameters.
fix: Modify a process that may cause a concurrent map write crash
feat: Bootstrap nodes no longer load and connect to history nodes
fix: ghw may encounter problems under macos
feat: Refactoring scheduled tasks
feat: Add a scheduled task for automatic upgrade
feat: The chat model supports image recognition
docs: Update documentations

v0.1.9

feat: Add an Idle value indicator to the model as an important condition for the proxy interface to filter nodes.
feat: Optimize the timeout.
feat: Modify the connection of the bootstrap node.
feat: The proxy interface requested using project name only sends a request once.
feat: Add some control interfaces for operating the bootstrap node.
feat: Add a configuration item called DialTimeout.
fix: Residual value when traversing leveldb.

v0.1.8

feat: The chat proxy interface no longer RoundTrip requests to the chat interface, but uses function reuse
feat: Replace net/http with the gin framework
feat: Modify some interfaces in accordance with the Restful HTTP API style

v0.1.7

feat: Restriction AI model and registered node must be on the same machine
fix: Fix the interface for calling the text-to-image model using the project name
fix: There may be crashes caused by null pointers
feat: The b64_json mode of the image is transmitted through a direct channel and is no longer broadcast to the distributed network
feat: Optimize the timeout processing of the text-to-image model

v0.1.6

feat: Improved error handing for chat model requests
feat: Rename the client/server node to the worker/input node
feat: Modify the image storage and transmission method of the text-to-image model
docs: Update documentations about text-to-image model

v0.1.5

fix: libp2p-ping stream cannot be released, the number has reached the upper limit
fix: Use parameters to pass context instead of saving it in a structure
feat: Add prometheus metrics processing
feat: Adjust the GracePeriod of the connection manager and the routing mode of pubsub
feat: Annotate the empty value judgment of wallet parameters in the AI model request

v0.1.4

Modification content:

1. Optimize the AI model call interface that uses the project name and without the node_id parameter.
2. Add wallet signature verification.

v0.1.3

Modification content:

1. The client node adds a dedicated project name configuration. Once set, it means that the node is dedicated to a certain AI project and will reject connections from other nodes that are not in this project.
2. Modify the client query interface for running specified AI projects and models, and add information such as RTT round-trip delay time in the interface response information.

v0.1.2

feat: Add some logs to chat model streaming