# Summary
Delete the "eval_results" directory, since it is already available in the training bucket: https://huggingface.co/buckets/rparkr/lfm-coder-training-bucket

# Motivation
The "eval_results" directory contained three JSONL files detailing the model's peformance on the evaluation datasets at three points during training (every 1,000 steps). Each file contains the prompt and model completion for all prompts across both evaluation datasets, the extracted code from the model's completion, and whether the model's code passed the tests. This data is useful for inspection and analysis, but it belongs in the training bucket with other training data (e.g., trackio metrics) rather than in the model repository.