Evaluation stops really early

[lesshaste]

When I run openevolve it all seems to be working fine but it just stops really early. The end of the output is:

2025-11-23 10:01:34,598 - INFO - Metrics: combined_score=-70.0000, reliability_score=1.0000, value_score=-70.0000
2025-11-23 10:01:34,780 - WARNING - Evaluation timed out after 60s
2025-11-23 10:01:34,788 - INFO - Received signal 15, initiating graceful shutdown...
2025-11-23 10:01:34,788 - INFO - Graceful shutdown requested...
2025-11-23 10:01:34,791 - INFO - Shutdown requested, canceling remaining evaluations...
2025-11-23 10:01:34,791 - INFO - ✅ Evolution completed - Shutdown requested
2025-11-23 10:01:34,791 - INFO - Evolution stopped due to shutdown request
2025-11-23 10:01:34,897 - INFO - Sampled model: openai/gpt-oss-120b
2025-11-23 10:01:35,007 - INFO - HTTP Request: POST https://openrouter.ai/api/v1/chat/completions "HTTP/1.1 200 OK"
2025-11-23 10:01:35,467 - INFO - HTTP Request: POST https://openrouter.ai/api/v1/chat/completions "HTTP/1.1 200 OK"
2025-11-23 10:01:39,181 - INFO - Evaluated program 3a5f0986-51e4-46f7-81dd-a6ef262a9a1e in 0.02s: combined_score=-10000.0000, reliability_score=0.0000, value_score=-10000.0000
2025-11-23 10:01:46,268 - INFO - Evaluated program c2c2fd92-f713-41e8-a443-7a6f6dce2a96 in 0.03s: combined_score=-10000.0000, reliability_score=0.0000, value_score=-10000.0000
2025-11-23 10:01:50,300 - WARNING - Evaluation timed out after 60s
2025-11-23 10:01:50,306 - INFO - Received signal 15, initiating graceful shutdown...
2025-11-23 10:01:50,306 - INFO - Graceful shutdown requested...
2025-11-23 10:01:50,527 - INFO - Evaluated program c44633b2-c804-4a8b-b172-735e238af69c in 0.03s: combined_score=-42.0000, reliability_score=1.0000, value_score=-42.0000
2025-11-23 10:02:13,113 - WARNING - Evaluation timed out after 60s
2025-11-23 10:02:13,118 - INFO - Received signal 15, initiating graceful shutdown...
2025-11-23 10:02:13,118 - INFO - Graceful shutdown requested...
2025-11-23 10:02:13,243 - INFO - Stopped process pool
2025-11-23 10:02:13,243 - INFO - Using tracked best program: 60c402bc-42e1-495b-b244-1e31b0700d1e
2025-11-23 10:02:13,243 - INFO - Evolution complete. Best program has metrics: combined_score=-42.0000, reliability_score=1.0000, value_score=-42.0000
2025-11-23 10:02:13,244 - INFO - Saved best program to examples/row_reduction/openevolve_output/best/best_program.py with program info to examples/row_reduction/openevolve_output/best/best_program_info.json

Evolution complete!
Best program metrics:
combined_score: -42.0000
reliability_score: 1.0000
value_score: -42.0000

Latest checkpoint saved at: examples/row_reduction/openevolve_output/checkpoints/checkpoint_20
To resume, use: --checkpoint examples/row_reduction/openevolve_output/checkpoints/checkpoint_20

My config.yml is:

Configuration for matrix example

max_iterations: 500
checkpoint_interval: 5

LLM configuration

llm:
primary_model: "openai/gpt-oss-120b"
primary_model_weight: 0.5
secondary_model: "x-ai/grok-4.1-fast"
secondary_model_weight: 0.5
api_base: "https://openrouter.ai/api/v1"
temperature: 0.7
max_tokens: 16000
timeout: 120

Prompt configuration

prompt:
system_message: PROMPT

Database configuration

database:
population_size: 70
archive_size: 20
num_islands: 6
elite_selection_ratio: 0.2
exploitation_ratio: 0.7

embedding_model: "text-embedding-3-small"
similarity_threshold: 0.00

Evaluator configuration

evaluator:
timeout: 60
parallel_evaluations: 5
cascade_evaluation: false

Evolution settings

diff_based_evolution: true
max_code_length: 20000

log_level: "INFO"

evolution_settings:
max_concurrent_generations: 1

What am I doing wrong?

[codelion]

2025-11-23 10:01:34,780 - WARNING - Evaluation timed out after 60s
2025-11-23 10:01:34,788 - INFO - Received signal 15, initiating graceful shutdown...
2025-11-23 10:01:34,788 - INFO - Graceful shutdown requested...
2025-11-23 10:01:34,791 - INFO - Shutdown requested, canceling remaining evaluations...

I think the evaluation is timing out and sending SIGTEM to the main process causing it to shutdown. Can you check your evaluator? Or increase the evaluation timeout to a larger value like 300 and try.

[lesshaste]

Thank you. I fixed the evaluator.