“config.yalm” and “config_planning.yalm” files are missing.

[zhang-he-666]

Hello, thank you for your hard work and dedication. However, when I tried to reproduce your code, I found that the "config.yalm" and "config_planning.yalm" files were missing. Could you help me with this?

[malong0115]

Same here

[malong0115]

Same here

[aTnT]

You were probably running on vLLM? vLLM primarily targets GPU acceleration and doesn't fully support CPU-only mode for inference, so it still requires a CUDA-capable GPU for execution.

[yangjian1218]

same problem。
Traceback (most recent call last):
File "/data1/code/yangjian/openAI/Paper2Code/scripts/../codes/3_coding_llm.py", line 55, in
with open(f'{output_dir}/planning_config.yaml') as f:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: '../outputs/Transformer_dscoder/planning_config.yaml'

[akbwork]

Is there a solution for this ?

[Tangyueming311]

same question

[fbrynpk]

I'm assuming people who's encounting this error is running the bash script for llm, which will run multiple python code consecutively, the config.yaml and planning_config.yaml file should be generated in the 1_planning_llm.py, if that fails, the rest of the script will fail as well, so the problem should be within 1_planning_llm.py this python files since it didn't generate the config file, for me this was the error

ValueError: The decoder prompt (length 21490) is longer than the maximum model length of 16384. Make sure that max_model_len is no smaller than the number of text tokens.
Traceback (most recent call last):
File "Paper2Code/scripts/../codes/1.1_extract_config.py", line 17, in
with open(f'{output_dir}/planning_trajectories.json', encoding='utf8') as f:
FileNotFoundError: [Errno 2] No such file or directory: '../outputs/Transformer_dscoder/planning_trajectories.json'
INFO 06-18 21:35:53 [init.py:244] Automatically detected platform cuda.
run_llm.sh: line 43: 1251 Segmentation fault (core dumped) python ../codes/2_analyzing_llm.py --paper_name $PAPER_NAME --model_name ${MODEL_NAME} --tp_size ${TP_SIZE} --pdf_json_path ${PDF_JSON_CLEANED_PATH} --output_dir ${OUTPUT_DIR}
INFO 06-18 21:36:03 [init.py:244] Automatically detected platform cuda.
Traceback (most recent call last):
File "Paper2Code/scripts/../codes/3_coding_llm.py", line 58, in
context_lst = extract_planning(f'{output_dir}/planning_trajectories.json')
File "Paper2Code/codes/utils.py", line 7, in extract_planning
with open(trajectories_json_file_path) as f:
FileNotFoundError: [Errno 2] No such file or directory: '../outputs/Transformer_dscoder/planning_trajectories.json'

The maximum model length was shorter than the total prompt when running the "DeepSeek-Coder-V2-Lite-Instruct" hence why it wasn't generating the config files for the next python scripts to use, I fixed it by running the model locally instead of from HF and it generates the config files accordingly