BamChatLLM doesn't correctly deserialize if it was originally created via BAMChatLLM.fromPreset

[shmuelk]

Describe the bug
When the class BAMChatLLM serializes itself, the createSnapShot function simply does a shallowCopy of the config.
This is problematic if the BAMChatLLM instance was created via BAMChatLLM.fromPreset, as the messagesToPrompt function is the loaded config references a local variable in the function in the class BAMChatLLMPreset.

I assume the same issue exists with the WatsonXChatLLM class.

To Reproduce
Steps to reproduce the behavior:

1. Use the following typescript file (it is the example examples/agent/bee_reusable.ts modified to work with BAM):

import "dotenv/config.js";
import { BeeAgent } from "bee-agent-framework/agents/bee/agent";
import { DuckDuckGoSearchTool } from "bee-agent-framework/tools/search/duckDuckGoSearch";
import { UnconstrainedMemory } from "bee-agent-framework/memory/unconstrainedMemory";
import { BAMChatLLM } from "bee-agent-framework/adapters/bam/chat";
 
const llm = BAMChatLLM.fromPreset("meta-llama/llama-3-8b-instruct");
    
// We create an agent
let agent = new BeeAgent({
  llm: llm,
  tools: [new DuckDuckGoSearchTool()],
  memory: new UnconstrainedMemory(),
});
  
// We ask the agent
let prompt = "Who is the president of USA?";
console.info(prompt);
const response = await agent.run({
  prompt,
});   
console.info(response.result.text);
  
// We can save (serialize) the agent
const json = agent.serialize();
    
// We reinitialize the agent to the exact state he was
agent = BeeAgent.fromSerialized(json);
  
// We continue in our conversation
prompt = "When was he born?";
console.info(prompt);
const response2 = await agent.run({
  prompt,
}); 
console.info(response2.result.text);

2. Set the GENAI_API_KEY environment variable to your BAM API key
3. Run: yarn start
4. See the error "template is not defined"

Expected behavior
The example should have worked normally.

Set-up:

• Bee version: [e.g. v0.0.27]
• Model provider [e.g. BAM]

[Tomas2D]

Thanks for the report. The issue has been fixed by d7584ef.

The fix will be released in 0.0.28 (expect a release later today or tomorrow).

[shmuelk]

Thanks I just pulled from main and the problem is fixed.

[Tomas2D]

Great. Released in v0.0.28.