[BUG] Human feedback errors when using training

[xpluscal]

Description

When trying to train, the following error occurs:

LiteLLM.Info: If you need to debug this error, use litellm.set_verbose=True'.

2025-01-09 14:07:54,742 - 8295578304 - llm.py-llm:187 - ERROR: LiteLLM call failed: litellm.BadRequestError: AnthropicException - Invalid first message=[]. Should always start with 'role'='user' for Anthropic. System prompt is sent separately for Anthropic. set 'litellm.modify_params = True' or 'litellm_settings:modify_params = True' on proxy, to insert a placeholder user message - '.' as the first message,
Received Messages=[]
Error during LLM call to classify human feedback: litellm.BadRequestError: AnthropicException - Invalid first message=[]. Should always start with 'role'='user' for Anthropic. System prompt is sent separately for Anthropic. set 'litellm.modify_params = True' or 'litellm_settings:modify_params = True' on proxy, to insert a placeholder user message - '.' as the first message,
Received Messages=[]. Retrying... (2/3)`

Steps to Reproduce

Create crew and run with train, provide feedback on 'looks good'

Expected behavior

Should store training

Screenshots/Code snippets

`from crewai import Agent, Task, Crew, LLM
from crewai_tools import (
DirectoryReadTool,
FileReadTool,
CodeInterpreterTool
)
from langchain_openai import ChatOpenAI

from dotenv import load_dotenv
import os

load_dotenv()

Create an agent with code execution enabled

coding_agent = Agent(
role="Python Data Analyst",
goal="Analyze the median Salary of participants. Must read the dataset.csv file from directory.",
backstory="You are an experienced data analyst with strong Python skills. Assume Python 3.12",
tools=[FileReadTool(), CodeInterpreterTool(), DirectoryReadTool(directory='./data')],
allow_code_execution=True,
max_retry_limit=3,
verbose=True,
llm=LLM(
model="anthropic/claude-3-5-sonnet-20240620",
temperature=0.1
),
)

Create a task that requires code execution

data_analysis_task = Task(
description="""
Analyze the dataset.csv file and calculate the median salary of participants.
Important: When using f-strings in Python code, ensure all quotes are properly escaped.
Use triple quotes for multi-line strings to avoid syntax errors.
""",
human_input=False,
expected_output="A markdown table showing the median salary of participants.",
agent=coding_agent
)

Create a crew with memory disabled to prevent loop persistence

analysis_crew = Crew(
agents=[coding_agent],
tasks=[data_analysis_task],
manager_llm=ChatOpenAI(model="gpt-4"),
verbose=True,
memory=True
)

Execute the crew

result = analysis_crew.train(n_iterations=3, filename='./data/training.pkl')
print(result)
`

Operating System

macOS Sonoma

Python Version

3.12

crewAI Version

0.95.0

crewAI Tools Version

0.25.8

Virtual Environment

Poetry

Evidence

Possible Solution

No idea.

Additional context