No such file or directory: 'templates/nanoGPT_lite/run_0/final_info.json'

[warys]

I used the local env and docker all showing flowing issue：

Response Content: {"total": 16, "offset": 0, "next": 10, "data": [{"paperId": "575506ce02dff846a9ec66a3c0cf69085228e435", "title": "Text Data Augmentation", "abstract": "Natural language processing model performance and generalization are greatly enhanced by text data augmentation. This paper introduces nlpaug, a Python library that provides a wide range of text data augmentation techniques. nlpaug offers functionalities for tasks such as synonym replacement, word insertion, word deletion, character-level modific Decision made: novel after round 3 Processing idea: adaptive_block_size Failed to evaluate idea adaptive_block_size: [Errno 2] No such file or directory: 'templates/nanoGPT_lite/run_0/final_info.json' Processing idea: layerwise_learning_rates Failed to evaluate idea layerwise_learning_rates: [Errno 2] No such file or directory: 'templates/nanoGPT_lite/run_0/final_info.json' Processing idea: data_augmentation Failed to evaluate idea data_augmentation: [Errno 2] No such file or directory: 'templates/nanoGPT_lite/run_0/final_info.json' All ideas evaluated.

which steps or missing on my env...

[conglu1997]

cd templates/nanoGPT_lite && python experiment.py --out_dir run_0 && python plot.py

Should be this line! :)

[NhanAyai]

I also encountered a similar issue: 'Traceback (most recent call last): File "/content/drive/MyDrive/AI-Scientist/templates/nanoGPT_lite/plot.py", line 15, in with open(osp.join(folder, "final_info.json"), "r") as f FileNotFoundError: [Errno 2] No such file or directory: 'run_0/final_info.json''. This is my code: '# NOTE: YOU MUST FIRST RUN THE PREPARE SCRIPTS ABOVE! %cd /content/drive/MyDrive/AI-Scientist/templates/nanoGPT_lite !python experiment.py --out_dir run_0 !python plot.py'. I have adjusted the data_dir to point to the correct location of the data in my directory, but the error still persists. I hope to receive some help, thank you.

[conglu1997]

Can you confirm that the final_info.json file has been created from the prep steps?

[axion15tth]

I faced the same problem before, but was able to solve it. In my case, the problem was still that cd templates/nanoGPT_lite && python experiment.py --out_dir run_0 && python plot.py was not executed correctly.

I missed it because there was no obvious error, but the GPU was not being used correctly and the training was not being done. I was using T4 and it didn't seem to be branching correctly, so I made the dtype explicit and it worked fine.

Before (experiment.py, line352)

dtype = (
        ‘bfloat16’.
        if torch.cuda.is_available() and torch.cuda.is_bf16_supported()
        else ‘float16’
    ) # ‘float32’, ‘bfloat16’, or ‘float16’, the latter will auto implement a GradScaler

After(experiment.py, line352) dtype = (‘float16’)

Then, line 374 was changed in the same way.