Nuitka: transformers, torch fail with triton installed, giving `OSError: could not get source code` in `triton` when importing `peft`

When importing peft executable is created, but when running it, It throws: OSError: could not get source code and exits.

Full log:

False
/tmp/onefile_10990_1700048433_451408/bitsandbytes/cextension.py:34: UserWarning: The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers, 8-bit m
ultiplication, and GPU quantization are unavailable.
'NoneType' object has no attribute 'cadam32bit_grad_fp32'
Traceback (most recent call last):
  File "/tmp/onefile_10990_1700048433_451408/triton_reproduce.py", line 7, in <module>
    from peft import AutoPeftModelForCausalLM, AutoPeftModelForSeq2SeqLM

  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/__init__.py", line 22, in <module peft>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/auto.py", line 30, in <module peft.auto>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/mapping.py", line 20, in <module peft.mapping>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/peft_model.py", line 39, in <module peft.peft_model>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/tuners/__init__.py", line 21, in <module peft.tuners>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/peft/tuners/lora.py", line 42, in <module peft.tuners.lora>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/bitsandbytes/__init__.py", line 16, in <module bitsandbytes>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/bitsandbytes/nn/__init__.py", line 6, in <module bitsandbytes.nn>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/bitsandbytes/nn/triton_based_modules.py", line 8, in <module bitsandbytes.nn.triton_based_modules>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/bitsandbytes/triton/dequantize_rowwise.py", line 10, in <module bitsandbytes.triton.dequantize_rowwise>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/triton/__init__.py", line 29, in <module triton>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/triton/language/__init__.py", line 8, in <module triton.language>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/triton/language/libdevice.py", line 4, in <module triton.language.libdevice>
  File "<frozen importlib._bootstrap>", line 1176, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1147, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
  File "/tmp/onefile_10990_1700048433_451408/triton/language/core.py", line 1175, in <module triton.language.core>
  File "/tmp/onefile_10990_1700048433_451408/triton/runtime/jit.py", line 419, in jit
  File "/tmp/onefile_10990_1700048433_451408/triton/runtime/jit.py", line 412, in decorator
  File "/tmp/onefile_10990_1700048433_451408/triton/runtime/jit.py", line 306, in __init__
  File "/tmp/onefile_10990_1700048433_451408/inspect.py", line 1262, in getsource
  File "/tmp/onefile_10990_1700048433_451408/inspect.py", line 1244, in getsourcelines
  File "/tmp/onefile_10990_1700048433_451408/inspect.py", line 1081, in findsource
OSError: could not get source code

I believe that the relevant source line is: https://github.com/openai/triton/blob/728282b11eebca01a64853d251b9fa9af0afe8f1/python/triton/runtime/jit.py#L585C44-L585C53

Reproduce details as guided:

• Nuitka version, full Python version, flavor, OS, etc. as output by this exact command.

1.9rc6
Commercial: 2.3.2
Python: 3.11.6 (main, Oct 19 2023, 12:55:47) [GCC 5.4.0 20160609]
Flavor: pyenv
Executable: /root/.pyenv/versions/3.11.6/bin/python
OS: Linux
Arch: x86_64
Distribution: Ubuntu (based on Debian) 16.04.7
Version C compiler: /usr/lib/ccache/gcc (gcc 5.4.0).

We must use staging version until some other issues are merged.

• How did you install Nuitka and Python pip install Nuitka-commercial-staging.zip no virtualenv.

The specific PyPI names and versions

``` accelerate==0.20.3 aiohttp==3.8.5 aiosignal==1.3.1 async-timeout==4.0.3 attrs==23.1.0 backoff==2.2.1 bitsandbytes==0.41.1 certifi==2023.7.22 charset-normalizer==3.2.0 click==8.1.7 cmake==3.27.7 datasets==2.14.5 Deprecated==1.2.14 dill==0.3.7 einops==0.6.1 filelock==3.12.4 frozenlist==1.4.0 fsspec==2023.6.0 googleapis-common-protos==1.60.0 grpc-interceptor==0.15.3 grpcio==1.58.0 grpcio-reflection==1.58.0 grpcio-status==1.58.0 grpcio-tools==1.51.1 hf_transfer==0.1.3 huggingface-hub==0.16.4 idna==3.4 Jinja2==3.1.2 lit==17.0.5 loguru==0.6.0 MarkupSafe==2.1.3 mpmath==1.3.0 multidict==6.0.4 multiprocess==0.70.15 mypy-protobuf==3.4.0 networkx==3.1 Nuitka @ file:///python-packages/Nuitka-commercial-staging.zip#sha256=77333731fa0274fbabcc4f7a959c57903a7bd800e563f761ed318f0c54676c5e numpy==1.26.0 nvidia-cublas-cu11==11.10.3.66 nvidia-cuda-cupti-cu11==11.7.101 nvidia-cuda-nvrtc-cu11==11.7.99 nvidia-cuda-runtime-cu11==11.7.99 nvidia-cudnn-cu11==8.5.0.96 nvidia-cufft-cu11==10.9.0.58 nvidia-curand-cu11==10.2.10.91 nvidia-cusolver-cu11==11.4.0.1 nvidia-cusparse-cu11==11.7.4.91 nvidia-nccl-cu11==2.14.3 nvidia-nvtx-cu11==11.7.91 opentelemetry-api==1.15.0 opentelemetry-exporter-otlp==1.15.0 opentelemetry-exporter-otlp-proto-grpc==1.15.0 opentelemetry-exporter-otlp-proto-http==1.15.0 opentelemetry-instrumentation==0.36b0 opentelemetry-instrumentation-grpc==0.36b0 opentelemetry-proto==1.15.0 opentelemetry-sdk==1.15.0 opentelemetry-semantic-conventions==0.36b0 ordered-set==4.1.0 packaging==23.1 pandas==2.1.1 peft==0.4.0 Pillow==10.0.1 protobuf==4.24.3 psutil==5.9.5 pyarrow==13.0.0 python-dateutil==2.8.2 pytz==2023.3.post1 PyYAML==6.0.1 regex==2023.8.8 requests==2.31.0 safetensors==0.3.3 scipy==1.11.2 sentencepiece==0.1.99 six==1.16.0 sympy==1.12 -e git+ssh://git@github.com/huggingface/text-generation-inference.git@a5def7c222174e03d815f890093584f3e815c5ce#egg=text_generation_server&subdirectory=server texttable==1.6.7 tokenizers==0.13.3 torch==2.0.1 tqdm==4.66.1 transformers==4.33.2 triton==2.0.0 typer==0.6.1 types-protobuf==4.24.0.4 typing_extensions==4.8.0 tzdata==2023.3 urllib3==2.0.5 wrapt==1.15.0 xxhash==3.3.0 yarl==1.9.2 zstandard==0.22.0 ```python

Other programs do work. even those which includes datasets and evaluate.

• Also supply a Short, Self Contained, Correct, Example

import os
import json
from loguru import logger
import torch

from transformers import AutoTokenizer
from peft import AutoPeftModelForCausalLM, AutoPeftModelForSeq2SeqLM


def download_and_unload_peft(model_id):
    torch_dtype = torch.float16

    logger.info("Peft model detected.")
    logger.info("Loading the model it might take a while without feedback")
    try:
        model = AutoPeftModelForCausalLM.from_pretrained(
            model_id,
            torch_dtype=torch_dtype,
            trust_remote_code=True,
            low_cpu_mem_usage=True,
        )
    except Exception:
        model = AutoPeftModelForSeq2SeqLM.from_pretrained(
            model_id,
            torch_dtype=torch_dtype,
            trust_remote_code=True,
            low_cpu_mem_usage=True,
        )
    logger.info("Loaded.")
    logger.info("Merging the lora weights.")

    base_model_id = model.peft_config["default"].base_model_name_or_path

    model = model.merge_and_unload()

    os.makedirs(model_id, exist_ok=True)
    cache_dir = model_id
    logger.info(f"Saving the newly created merged model to {cache_dir}")
    tokenizer = AutoTokenizer.from_pretrained(base_model_id)
    model.save_pretrained(cache_dir, safe_serialization=True)
    model.config.save_pretrained(cache_dir)
    tokenizer.save_pretrained(cache_dir)


download_and_unload_peft("ybelkada/opt-350m-lora")

• Provide in your issue the Nuitka options used

python -m nuitka --main=triton_reproduce.py --standalone --onefile --static-libpython=no --include-distribution-met
adata=accelerate --include-distribution-metadata=bitsandbytes --include-distribution-metadata=datasets --include-distribution-metadata=jinja2 --include-distribution-metadata=pand
as --include-distribution-metadata=psutil --include-distribution-metadata=safetensors --include-distribution-metadata=torch --include-distribution-metadata=tqdm --include-distrib
ution-metadata=transformers --include-distribution-metadata=huggingface-hub --include-distribution-metadata=scipy --include-distribution-metadata=tokenizers --include-distributio
n-metadata=protobuf --include-distribution-metadata=peft --noinclude-unittest-mode=allow