Respan Dataset Explorer

Select one behavior. Every returned turn has one binary label: Present or Absent. Source: final dense boolean release.

5,167,182physical rows

86shards

0.00%qualified row coverage

0.00%qualified cell coverage

Shard Start offset Rows Behavior Label Random row JSON API

turns-00078.parquet:16163

f562cf03612f0f6eb646ab5a

turn 1/1gpt-4.1-mini-2025-04-14EnglishSouth Korea219 words

degenerate_repetitionAbsentFinal dense release

USER

I have a list of ingredients that I would like to format in a simple and consistent way, following this structure:

"Quantity Unit Description."

Formatting Rules:

Quantity must be a number (e.g., "200" instead of "two hundred").

Unit must use the metric system and the corresponding abbreviated form (e.g., grams → g, liters → l, milliliters → ml, etc.).

Description should be the ingredient, including any specific details if mentioned (e.g., "brown sugar", "melted unsalted butter").

If an ingredient is already formatted correctly, do not change it.

Do not invent a quantity or unit if none is provided — leave it as is and just include the description.

If a line includes alternatives or multiple options (e.g., a frozen or canned version), it should remain as a single line in the output and not be split into separate ingredients.

Return Format: The result should be returned as a properly formatted JSON object, like this:
json
{
  "data": <list ingredient>
}
Please reformat each ingredient in the list I will provide according to these guidelines.

Context Dump
Ingredients source provided:
1 kg chicken livers,400 g butter,5 cl cognac,5 cl truffle juice,2 pinches allspice,salt, pepper

ASSISTANT

```json
{
  "data": [
    "1 kg chicken livers.",
    "400 g butter.",
    "5 cl cognac.",
    "5 cl truffle juice.",
    "2 pinches allspice.",
    "salt.",
    "pepper."
  ]
}
```

turns-00078.parquet:16164

17cb5402a0d0b91fa6dee8bc

turn 1/1gpt-4.1-mini-2025-04-14EnglishUnited States204 words

degenerate_repetitionAbsentFinal dense release

USER

I have a list of ingredients that I would like to format in a simple and consistent way, following this structure:

"Quantity Unit Description."

Formatting Rules:

Quantity must be a number (e.g., "200" instead of "two hundred").

Unit must use the metric system and the corresponding abbreviated form (e.g., grams → g, liters → l, milliliters → ml, etc.).

Description should be the ingredient, including any specific details if mentioned (e.g., "brown sugar", "melted unsalted butter").

If an ingredient is already formatted correctly, do not change it.

Do not invent a quantity or unit if none is provided — leave it as is and just include the description.

If a line includes alternatives or multiple options (e.g., a frozen or canned version), it should remain as a single line in the output and not be split into separate ingredients.

Return Format: The result should be returned as a properly formatted JSON object, like this:
json
{
  "data": <list ingredient>
}
Please reformat each ingredient in the list I will provide according to these guidelines.

Context Dump
Ingredients source provided:
50 cl dry white wine,1 bouquet garni,1 onion

ASSISTANT

```json
{
  "data": [
    "500 ml dry white wine.",
    "1 bouquet garni.",
    "1 onion."
  ]
}
```

turns-00078.parquet:16165

17cb5402a0d0b91fa6dee8bc

turn 1/1gpt-4.1-mini-2025-04-14Englishunknown country204 words

degenerate_repetitionAbsentFinal dense release

USER

I have a list of ingredients that I would like to format in a simple and consistent way, following this structure:

"Quantity Unit Description."

Formatting Rules:

Quantity must be a number (e.g., "200" instead of "two hundred").

Unit must use the metric system and the corresponding abbreviated form (e.g., grams → g, liters → l, milliliters → ml, etc.).

Description should be the ingredient, including any specific details if mentioned (e.g., "brown sugar", "melted unsalted butter").

If an ingredient is already formatted correctly, do not change it.

Do not invent a quantity or unit if none is provided — leave it as is and just include the description.

If a line includes alternatives or multiple options (e.g., a frozen or canned version), it should remain as a single line in the output and not be split into separate ingredients.

Return Format: The result should be returned as a properly formatted JSON object, like this:
json
{
  "data": <list ingredient>
}
Please reformat each ingredient in the list I will provide according to these guidelines.

Context Dump
Ingredients source provided:
50 cl dry white wine,1 bouquet garni,1 onion

ASSISTANT

```json
{
  "data": [
    "500 ml dry white wine.",
    "1 bouquet garni.",
    "1 onion."
  ]
}
```

turns-00078.parquet:16166

b34dc749d8040f434c970d9e

turn 1/1gpt-4.1-mini-2025-04-14ArabicChina115 words

degenerate_repetitionAbsentFinal dense release

USER

ماهي الثغرة

ASSISTANT

الثغرة مصطلح يُستخدم في مجال أمن المعلومات ويشير إلى نقطة ضعف أو خلل في نظام برمجي أو شبكة حاسوبية يمكن للمهاجمين استغلالها للوصول غير المصرح به أو تنفيذ هجمات تؤدي إلى أضرار مثل سرقة البيانات، تعطيل الخدمة، أو السيطرة على الأنظمة.

بشكل عام، يمكن تلخيص معنى الثغرة في:

- **ثغرة أمنية**: هو خلل في البرمجيات أو إعدادات النظام يمكن استغلاله لاختراق النظام.
- **أنواع الثغرات**: تشمل ثغرات في البرمجيات مثل ثغرات SQL Injection، XSS، Buffer Overflow، أو في الشبكات مثل ضعف في إعدادات الجدار الناري.
- **أهمية اكتشاف الثغرات**: مهم لتأمين الأنظمة عبر تحديث البرمجيات، استخدام أدوات الحماية، وتصحيح العيوب فور اكتشافها.

هل تود معرفة المزيد عن أنواع الثغرات أو طرق الوقاية منها؟

turns-00078.parquet:16167

14f16c269a0a475e5c9e61af

turn 1/1gpt-4.1-mini-2025-04-14Arabicunknown country94 words

degenerate_repetitionAbsentFinal dense release

USER

صابونة Baby Soft للبشرة الجافة

ASSISTANT

صابونة Baby Soft للبشرة الجافة هي صابونة مصممة خصيصًا لترطيب وتنظيف البشرة الحساسة والجافة بلطف. تحتوي عادةً على مكونات مرطبة تساعد في الحفاظ على نعومة البشرة وتمنع جفافها، مثل زيت الزيتون، الجليسرين، أو زبدة الشيا.

إذا كنت تبحث عن استخدام هذه الصابونة، فمن الأفضل التأكد من قراءة المكونات على العبوة للتأكد من أنها مناسبة لبشرتك وعدم وجود أي مواد قد تسبب لك حساسية. كما يُفضل استخدامها مع مرطب بعد الاستحمام للحفاظ على ترطيب البشرة.

هل ترغب في معرفة أماكن شراء صابونة Baby Soft أو نصائح إضافية للعناية بالبشرة الجافة؟

turns-00078.parquet:16168

63650dd97169c16fe03e45d1

turn 1/1gpt-4.1-mini-2025-04-14EnglishChina23 words

degenerate_repetitionAbsentFinal dense release

USER

User: hi, how are you?
User: hi, how are you?
Assistant:

ASSISTANT

Hello! I'm doing well, thank you. How can I assist you today?

turns-00078.parquet:16169

4208da3f80e0932df1f3cc48

turn 1/6gpt-4.1-mini-2025-04-14EnglishUnited States584 words

degenerate_repetitionAbsentFinal dense release

USER

(ace_step) C:\AI\ACE-Step>python trainer.py --dataset_path "C:\Users\DjZaN\OneDrive\Desktop\Zen\dataset.json" --checkpoint_dir "C:\Users\DjZaN\.cache\ace-step\checkpoints" --exp_name "DaltOniX"
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\requests\__init__.py:86: RequestsDependencyWarning: Unable to find acceptable character detection dependency (chardet or charset_normalizer).
  warnings.warn(
[2025-05-09 16:14:39,198] [INFO] [real_accelerator.py:158:get_accelerator] Setting ds_accelerator to cuda (auto detect)
W0509 16:14:39.808000 11976 site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs.
Traceback (most recent call last):
  File "C:\AI\ACE-Step\trainer.py", line 861, in <module>
    main(args)
  File "C:\AI\ACE-Step\trainer.py", line 813, in main
    trainer = Trainer(
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\pytorch_lightning\utilities\argparse.py", line 70, in insert_env_defaults
    return fn(self, **kwargs)
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 404, in __init__
    self._accelerator_connector = _AcceleratorConnector(
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\pytorch_lightning\trainer\connectors\accelerator_connector.py", line 157, in __init__
    self._init_strategy()
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\pytorch_lightning\trainer\connectors\accelerator_connector.py", line 480, in _init_strategy
    self.strategy = StrategyRegistry.get(self._strategy_flag)
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\lightning_fabric\strategies\registry.py", line 94, in get
    return data["strategy"](**data["init_params"])
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\pytorch_lightning\strategies\deepspeed.py", line 307, in __init__
    import deepspeed
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\__init__.py", line 22, in <module>
    from . import module_inject
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\module_inject\__init__.py", line 6, in <module>
    from .replace_module import replace_transformer_layer, revert_transformer_layer, ReplaceWithTensorSlicing, GroupQuantizer, generic_injection
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\module_inject\replace_module.py", line 587, in <module>
    from ..pipe import PipelineModule
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\pipe\__init__.py", line 6, in <module>
    from ..runtime.pipe import PipelineModule, LayerSpec, TiedLayerSpec
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\runtime\pipe\__init__.py", line 6, in <module>
    from .module import PipelineModule, LayerSpec, TiedLayerSpec
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\runtime\pipe\module.py", line 19, in <module>
    from ..activation_checkpointing import checkpointing
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\runtime\activation_checkpointing\checkpointing.py", line 26, in <module>
    from deepspeed.runtime.config import DeepSpeedConfig
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\runtime\config.py", line 41, in <module>
    from ..elasticity import (
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\elasticity\__init__.py", line 10, in <module>
    from .elastic_agent import DSElasticAgent
  File "C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\deepspeed\elasticity\elastic_agent.py", line 9, in <module>
    from torch.distributed.elastic.agent.server.api import log, _get_socket_with_port
ImportError: cannot import name 'log' from 'torch.distributed.elastic.agent.server.api' (C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\distributed\elastic\agent\server\api.py)

ASSISTANT

The error message you're encountering:

```
ImportError: cannot import name 'log' from 'torch.distributed.elastic.agent.server.api'
```

comes from the `deepspeed` package trying to import `log` from `torch.distributed.elastic.agent.server.api`. This indicates an incompatibility between your installed PyTorch version and the `deepspeed` version you're using.

---

### Root cause:
PyTorch recently refactored or changed its elastic/distributed API, removing or renaming some imports. Your installed `deepspeed` version is expecting an older API version where `log` existed inside `torch.distributed.elastic.agent.server.api`, but your current PyTorch version no longer provides it.

---

### How to fix:

#### 1. Check your PyTorch version
Run:
```bash
python -c "import torch; print(torch.__version__)"
```
Be aware of the major version you're using.

#### 2. Update or downgrade DeepSpeed
- **Update DeepSpeed** to the latest version, which may have fixed this compatibility issue:
```bash
pip install --upgrade deepspeed
```
- If the latest DeepSpeed does not support your PyTorch version or it's still broken, try downgrading PyTorch to a version compatible with your DeepSpeed install, for example PyTorch 1.13.x if you're currently on 2.x.

Common compatible versions:
- DeepSpeed v0.9.x works well with PyTorch 1.13.x
- Newer DeepSpeed versions (>=0.10) might be needed for PyTorch 2.x

#### 3. Optionally, pin a working combination (example):

```bash
pip install torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu117
pip install deepspeed==0.9.2
```

(adjust CUDA version as appropriate)

#### 4. Alternatively, patch or work around
If updating/downgrading is not immediately possible, as a last resort, you could:
- Fork DeepSpeed and comment out or fix the import.
- Or, try forcing the Lightning Trainer to use a different `strategy` than DeepSpeed if you don't explicitly need deepspeed features.

E.g., change your trainer initialization or add a CLI flag to avoid defaulting to DeepSpeed if possible:
```python
Trainer(strategy=None)  # or "ddp" or "ddp_spawn" or "auto"
```

---

### Summary

Your main action items are:

- Ensure your torch and deepspeed versions are compatible.
- Upgrade `deepspeed` to latest version.
- If problem persists, downgrade PyTorch or change your training strategy.

---

If you want, provide the output of:

```bash
pip show torch
pip show deepspeed
```

I can help suggest exact version combos.

turns-00078.parquet:16170

5d1d9e2bb85c57643a01e0a0

turn 2/6gpt-4.1-mini-2025-04-14EnglishUnited States496 words

degenerate_repetitionAbsentFinal dense release

USER

how to compile deepspeed for windows

ASSISTANT

DeepSpeed does **not officially support Windows** natively, and compiling it on Windows is challenging because it relies heavily on Linux-specific tools and libraries. This is why you often encounter issues trying to run DeepSpeed on Windows environments.

However, you do have a few options and workarounds:

---

### Official Status
- DeepSpeed is primarily designed for Linux.
- Microsoft and the DeepSpeed team do not provide official Windows binaries.
- Building DeepSpeed on Windows requires custom setups and sometimes cannot work perfectly.

---

### Ways to Use or Compile DeepSpeed on Windows

#### 1. **Use WSL2 (Windows Subsystem for Linux v2)**
The best and most practical solution is to install and run DeepSpeed inside a Linux environment on Windows, provided by WSL2.

**Steps:**

- Install WSL2 with a Linux distro (Ubuntu recommended):

```powershell
wsl --install
```

- Open your Ubuntu shell via WSL.
- Install CUDA for WSL if you have an NVIDIA GPU (instructions from NVIDIA).
- Install PyTorch and DeepSpeed inside the WSL environment using Linux instructions:

```bash
pip install torch torchvision
pip install deepspeed
```

- Then run training scripts inside WSL terminal.

This method gives you near-native Linux GPU support on Windows without dual boot or separate machine.

---

#### 2. **Build DeepSpeed on Windows (Not recommended / advanced)**

If you *really* want to try compiling DeepSpeed directly on Windows (not guaranteed to work fully):

- Install **Visual Studio with C++ Build Tools**.

- Install **CUDA Toolkit** for Windows.

- Make sure you have Python, pip, and required dependencies.

- Clone the DeepSpeed repo:

```bash
git clone https://github.com/microsoft/DeepSpeed.git
cd DeepSpeed
```

- DeepSpeed uses custom CUDA extensions, so try to build using:

```bash
DS_BUILD_CUDA=1 pip install .
```

- You may run into errors related to `setup.py` and CUDA extensions that expect Linux paths, commands (`make`), or features.

- You'd need to manually fix C++/CUDA build scripts to adapt to Windows paths and tooling.

**Practically, this is difficult and error prone. No official docs exist for Windows.**

---

#### 3. **Docker with Linux Container on Windows**
If you prefer, you can:

- Install Docker Desktop with WSL2 backend enabled.

- Use a Linux container with CUDA support (NVIDIA Container Toolkit).

- Build and run DeepSpeed inside the container.

This is often easier than building natively on Windows.

---

### Summary
| Option | Pros | Cons |
|-------------------------|---------------------------------|---------------------------------|
| WSL2 | Easy, official Linux support | Slight overhead, need WSL2 setup |
| Native Windows build | (None really, experimental) | Complex, often broken |
| Docker (Linux container) | Clean isolated Linux environment | Requires Docker setup |

---

### Recommendation
For Windows users who want to run or compile DeepSpeed, **WSL2 is the recommended and easiest approach**.

---

### Additional Resources:

- [DeepSpeed GitHub Issues about Windows](https://github.com/microsoft/DeepSpeed/issues?q=windows)
- [WSL2 Installation guide](https://learn.microsoft.com/en-us/windows/wsl/install)
- [NVIDIA CUDA on WSL](https://developer.nvidia.com/cuda/wsl)

---

If you want, I can help you set up WSL2 + DeepSpeed step-by-step. Would you like that?

turns-00078.parquet:16171

832dcea7f1e9a94646004acd

turn 3/6gpt-4.1-mini-2025-04-14EnglishUnited States6021 words

degenerate_repetitionAbsentFinal dense release

USER

(ace_step) C:\AI\ACE-Step\DeepSpeed>"C:\AI\ACE-Step\DeepSpeed\build_win.bat"
* Getting build dependencies for wheel...
DS_BUILD_OPS=1
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Install Ops={'async_io': False, 'fused_adam': 1, 'cpu_adam': 1, 'cpu_adagrad': 1, 'cpu_lion': 1, 'dc': False, 'evoformer_attn': False, 'fp_quantizer': False, 'fused_lamb': 1, 'fused_lion': 1, 'gds': False, 'transformer_inference': 1, 'inference_core_ops': 1, 'cutlass_ops': False, 'quantizer': 1, 'ragged_device_ops': False, 'ragged_ops': 1, 'random_ltd': 1, 'sparse_attn': False, 'spatial_inference': 1, 'transformer': 1, 'stochastic_transformer': 1}
version=0.16.8+unknown, git_hash=unknown, git_branch=unknown
install_requires=['einops', 'hjson', 'msgpack', 'ninja', 'numpy', 'packaging>=20.0', 'psutil', 'py-cpuinfo', 'pydantic>=2.0.0', 'torch', 'tqdm', 'nvidia-ml-py']
ext_modules=[<setuptools.extension.Extension('deepspeed.ops.adam.fused_adam_op') at 0x26f48f93a60>, <setuptools.extension.Extension('deepspeed.ops.adam.cpu_adam_op') at 0x26f79a6ce80>, <setuptools.extension.Extension('deepspeed.ops.adagrad.cpu_adagrad_op') at 0x26f79a6d090>, <setuptools.extension.Extension('deepspeed.ops.lion.cpu_lion_op') at 0x26f79a6c4f0>, <setuptools.extension.Extension('deepspeed.ops.lamb.fused_lamb_op') at 0x26f79a6c190>, <setuptools.extension.Extension('deepspeed.ops.lion.fused_lion_op') at 0x26f79a6ce20>, <setuptools.extension.Extension('deepspeed.ops.transformer.inference.transformer_inference_op') at 0x26f79a6cf10>, <setuptools.extension.Extension('deepspeed.inference.v2.kernelsinference_core_ops') at 0x26f79a6cdf0>, <setuptools.extension.Extension('deepspeed.ops.quantizer.quantizer_op') at 0x26f79a6c490>, <setuptools.extension.Extension('deepspeed.inference.v2.ragged_ops') at 0x26f79a6d150>, <setuptools.extension.Extension('deepspeed.ops.random_ltd_op') at 0x26f79a6cfd0>, <setuptools.extension.Extension('deepspeed.ops.spatial.spatial_inference_op') at 0x26f79a6cee0>, <setuptools.extension.Extension('deepspeed.ops.transformer.transformer_op') at 0x26f79a6cfa0>, <setuptools.extension.Extension('deepspeed.ops.transformer.stochastic_transformer_op') at 0x26f79a6cca0>]
running egg_info
creating deepspeed.egg-info
writing deepspeed.egg-info\PKG-INFO
writing dependency_links to deepspeed.egg-info\dependency_links.txt
writing entry points to deepspeed.egg-info\entry_points.txt
writing requirements to deepspeed.egg-info\requires.txt
writing top-level names to deepspeed.egg-info\top_level.txt
writing manifest file 'deepspeed.egg-info\SOURCES.txt'
reading manifest file 'deepspeed.egg-info\SOURCES.txt'
reading manifest template 'MANIFEST_win.in'
warning: no previously-included files matching '*.cpp' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.h' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cu' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cuh' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cc' found under directory 'deepspeed\ops\csrc'
no previously-included directories found matching 'csrc'
no previously-included directories found matching 'op_builder'
no previously-included directories found matching 'accelerator'
adding license file 'LICENSE'
writing manifest file 'deepspeed.egg-info\SOURCES.txt'
deepspeed build time = 0.5883395671844482 secs
* Building wheel...
[2025-05-09 16:24:37,929] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-09 16:24:38,734] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
W0509 16:24:44.524000 22116 site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs.
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\requests\__init__.py:86: RequestsDependencyWarning: Unable to find acceptable character detection dependency (chardet or charset_normalizer).
  warnings.warn(
DS_BUILD_OPS=1
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
Install Ops={'async_io': False, 'fused_adam': 1, 'cpu_adam': 1, 'cpu_adagrad': 1, 'cpu_lion': 1, 'dc': False, 'evoformer_attn': False, 'fp_quantizer': False, 'fused_lamb': 1, 'fused_lion': 1, 'gds': False, 'transformer_inference': 1, 'inference_core_ops': 1, 'cutlass_ops': False, 'quantizer': 1, 'ragged_device_ops': False, 'ragged_ops': 1, 'random_ltd': 1, 'sparse_attn': False, 'spatial_inference': 1, 'transformer': 1, 'stochastic_transformer': 1}
version=0.16.8+unknown, git_hash=unknown, git_branch=unknown
install_requires=['einops', 'hjson', 'msgpack', 'ninja', 'numpy', 'packaging>=20.0', 'psutil', 'py-cpuinfo', 'pydantic>=2.0.0', 'torch', 'tqdm', 'nvidia-ml-py']
ext_modules=[<setuptools.extension.Extension('deepspeed.ops.adam.fused_adam_op') at 0x2029dc0f5e0>, <setuptools.extension.Extension('deepspeed.ops.adam.cpu_adam_op') at 0x202e8573be0>, <setuptools.extension.Extension('deepspeed.ops.adagrad.cpu_adagrad_op') at 0x202e8573c10>, <setuptools.extension.Extension('deepspeed.ops.lion.cpu_lion_op') at 0x2029de243d0>, <setuptools.extension.Extension('deepspeed.ops.lamb.fused_lamb_op') at 0x2029de246d0>, <setuptools.extension.Extension('deepspeed.ops.lion.fused_lion_op') at 0x2029de245b0>, <setuptools.extension.Extension('deepspeed.ops.transformer.inference.transformer_inference_op') at 0x2029e2d3f10>, <setuptools.extension.Extension('deepspeed.inference.v2.kernelsinference_core_ops') at 0x2029e2d3e50>, <setuptools.extension.Extension('deepspeed.ops.quantizer.quantizer_op') at 0x2029dd64430>, <setuptools.extension.Extension('deepspeed.inference.v2.ragged_ops') at 0x2029dd64f40>, <setuptools.extension.Extension('deepspeed.ops.random_ltd_op') at 0x2029dd64f70>, <setuptools.extension.Extension('deepspeed.ops.spatial.spatial_inference_op') at 0x2029dd644f0>, <setuptools.extension.Extension('deepspeed.ops.transformer.transformer_op') at 0x2029dd64730>, <setuptools.extension.Extension('deepspeed.ops.transformer.stochastic_transformer_op') at 0x2029dd641f0>]
running bdist_wheel
running build
running build_py
creating build\lib.win-amd64-cpython-310\deepspeed
copying deepspeed\constants.py -> build\lib.win-amd64-cpython-310\deepspeed
copying deepspeed\env_report.py -> build\lib.win-amd64-cpython-310\deepspeed
copying deepspeed\git_version_info.py -> build\lib.win-amd64-cpython-310\deepspeed
copying deepspeed\git_version_info_installed.py -> build\lib.win-amd64-cpython-310\deepspeed
copying deepspeed\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed
creating build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\abstract_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\cpu_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\cuda_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\hpu_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\mlu_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\mps_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\npu_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\real_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\sdaa_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\xpu_accelerator.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
copying deepspeed\accelerator\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\accelerator
creating build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\autotuner.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\config.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\scheduler.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
copying deepspeed\autotuning\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning
creating build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\deepspeed_checkpoint.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\ds_to_universal.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\reshape_3d_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\reshape_meg_2d.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\reshape_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\universal_checkpoint.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\zero_checkpoint.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
copying deepspeed\checkpoint\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\checkpoint
creating build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\backend.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\ccl.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\comm.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\config.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\reduce_op.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\torch.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
copying deepspeed\comm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\comm
creating build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\backend.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\config.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\fx.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\graph_param.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\inductor.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\init_z1.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\init_z3.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\list_schedule.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\partitioner.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\patch_compiled_func.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\patch_fake_tensor.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\util.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
copying deepspeed\compile\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\compile
creating build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\basic_layer.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\compress.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\config.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\helper.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\scheduler.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
copying deepspeed\compression\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\compression
creating build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\config.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\elasticity.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\elastic_agent.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
copying deepspeed\elasticity\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\elasticity
creating build\lib.win-amd64-cpython-310\deepspeed\inference
copying deepspeed\inference\config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference
copying deepspeed\inference\engine.py -> build\lib.win-amd64-cpython-310\deepspeed\inference
copying deepspeed\inference\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference
creating build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\launch.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\launcher_helper.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\multinode_runner.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\runner.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
copying deepspeed\launcher\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\launcher
creating build\lib.win-amd64-cpython-310\deepspeed\linear
copying deepspeed\linear\config.py -> build\lib.win-amd64-cpython-310\deepspeed\linear
copying deepspeed\linear\context_manager.py -> build\lib.win-amd64-cpython-310\deepspeed\linear
copying deepspeed\linear\optimized_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\linear
copying deepspeed\linear\quantization.py -> build\lib.win-amd64-cpython-310\deepspeed\linear
copying deepspeed\linear\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\linear
creating build\lib.win-amd64-cpython-310\deepspeed\model_implementations
copying deepspeed\model_implementations\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations
creating build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\auto_tp.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\auto_tp_model_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\fusedqkv_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\inject.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\layers.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\load_checkpoint.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\module_quantize.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\replace_module.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\replace_policy.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\tp_shard.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
copying deepspeed\module_inject\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject
creating build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\experts.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\layer.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\mappings.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\sharded_moe.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
copying deepspeed\moe\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\moe
creating build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\comet.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\config.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\csv_monitor.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\monitor.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\tensorboard.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\wandb.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
copying deepspeed\monitor\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\monitor
creating build\lib.win-amd64-cpython-310\deepspeed\nebula
copying deepspeed\nebula\config.py -> build\lib.win-amd64-cpython-310\deepspeed\nebula
copying deepspeed\nebula\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\nebula
copying deepspeed\nebula\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\nebula
creating build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\ds_aio_args.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\ds_aio_basic.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\ds_aio_handle.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\ds_aio_job.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\parse_nvme_stats.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\perf_generate_param.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\perf_run_sweep.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\perf_sweep_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\test_ds_aio.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\test_ds_aio_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\validate_async_io.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
copying deepspeed\nvme\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\nvme
creating build\lib.win-amd64-cpython-310\deepspeed\ops
copying deepspeed\ops\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops
creating build\lib.win-amd64-cpython-310\deepspeed\pipe
copying deepspeed\pipe\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\pipe
creating build\lib.win-amd64-cpython-310\deepspeed\profiling
copying deepspeed\profiling\config.py -> build\lib.win-amd64-cpython-310\deepspeed\profiling
copying deepspeed\profiling\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\profiling
copying deepspeed\profiling\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\profiling
creating build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\base_optimizer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\bf16_optimizer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\compiler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\config_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\dataloader.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\eigenvalue.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\hybrid_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\lr_schedules.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\progressive_layer_drop.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\quantize.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\sparse_tensor.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\state_dict_factory.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\weight_quantizer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
copying deepspeed\runtime\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime
creating build\lib.win-amd64-cpython-310\deepspeed\sequence
copying deepspeed\sequence\cross_entropy.py -> build\lib.win-amd64-cpython-310\deepspeed\sequence
copying deepspeed\sequence\fpdt_layer.py -> build\lib.win-amd64-cpython-310\deepspeed\sequence
copying deepspeed\sequence\layer.py -> build\lib.win-amd64-cpython-310\deepspeed\sequence
copying deepspeed\sequence\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\sequence
creating build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\bwc.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\comms_logging.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\config.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\debug.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\exceptions.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\groups.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\init_on_device.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\logging.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\mixed_precision_linkage.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\numa.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\nvtx.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\tensor_fragment.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\timer.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\torch.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\types.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\z3_leaf_module.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\zero_to_fp32.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
copying deepspeed\utils\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\utils
creating build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\base_tuner.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\cost_model.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\index_based_tuner.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\model_based_tuner.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
copying deepspeed\autotuning\tuner\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\autotuning\tuner
creating build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\offload_activation.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\offload_adam_states.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\offload_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\prefetch.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\selective_gather.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\zero1_compile.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\zero3_compile.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
copying deepspeed\compile\passes\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\passes
creating build\lib.win-amd64-cpython-310\deepspeed\compile\profilers
copying deepspeed\compile\profilers\comm_profile.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\profilers
copying deepspeed\compile\profilers\graph_profile.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\profilers
copying deepspeed\compile\profilers\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\compile\profilers
creating build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
copying deepspeed\inference\quantization\layers.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
copying deepspeed\inference\quantization\quantization.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
copying deepspeed\inference\quantization\quantization_context.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
copying deepspeed\inference\quantization\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
copying deepspeed\inference\quantization\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\quantization
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\allocator.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\config_v2.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\engine_factory.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\engine_v2.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\inference_parameter.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\inference_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\logging.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\scheduling_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
copying deepspeed\inference\v2\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\checkpoint
copying deepspeed\inference\v2\checkpoint\base_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\checkpoint
copying deepspeed\inference\v2\checkpoint\huggingface_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\checkpoint
copying deepspeed\inference\v2\checkpoint\in_memory_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\checkpoint
copying deepspeed\inference\v2\checkpoint\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\checkpoint
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels
copying deepspeed\inference\v2\kernels\ds_kernel.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels
copying deepspeed\inference\v2\kernels\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\flat_model_helpers.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\inference_model_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\inference_policy_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\inference_transformer_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\layer_container_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\parameter_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
copying deepspeed\inference\v2\model_implementations\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules
copying deepspeed\inference\v2\modules\ds_module.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules
copying deepspeed\inference\v2\modules\heuristics.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules
copying deepspeed\inference\v2\modules\module_registry.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules
copying deepspeed\inference\v2\modules\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\blocked_allocator.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\kv_cache.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\manager_configs.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\ragged_manager.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\ragged_wrapper.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\sequence_descriptor.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
copying deepspeed\inference\v2\ragged\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops
copying deepspeed\inference\v2\kernels\core_ops\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops
copying deepspeed\inference\v2\kernels\cutlass_ops\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops
copying deepspeed\inference\v2\kernels\ragged_ops\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\bias_activation.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\blas_kernels
copying deepspeed\inference\v2\kernels\core_ops\blas_kernels\blas_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\blas_kernels
copying deepspeed\inference\v2\kernels\core_ops\blas_kernels\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\blas_kernels
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\cuda_fp_ln_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\cuda_ln.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\cuda_post_ln.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\cuda_pre_ln.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\cuda_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_pre_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\gated_activation.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\mixed_gemm
copying deepspeed\inference\v2\kernels\cutlass_ops\mixed_gemm\mixed_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\mixed_gemm
copying deepspeed\inference\v2\kernels\cutlass_ops\mixed_gemm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\mixed_gemm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm
copying deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm\mixed_moe_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm
copying deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm\moe_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm
copying deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\cutlass_ops\moe_gemm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\atom_builder
copying deepspeed\inference\v2\kernels\ragged_ops\atom_builder\atom_builder.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\atom_builder
copying deepspeed\inference\v2\kernels\ragged_ops\atom_builder\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\atom_builder
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\blocked_flash
copying deepspeed\inference\v2\kernels\ragged_ops\blocked_flash\blocked_flash.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\blocked_flash
copying deepspeed\inference\v2\kernels\ragged_ops\blocked_flash\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\blocked_flash
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\embed
copying deepspeed\inference\v2\kernels\ragged_ops\embed\embed.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\embed
copying deepspeed\inference\v2\kernels\ragged_ops\embed\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\embed
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary
copying deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary\blocked_kv_rotary.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary
copying deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary\blocked_trained_kv_rotary.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary
copying deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary\linear_blocked_kv_copy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary
copying deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\linear_blocked_kv_rotary
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\logits_gather
copying deepspeed\inference\v2\kernels\ragged_ops\logits_gather\logits_gather.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\logits_gather
copying deepspeed\inference\v2\kernels\ragged_ops\logits_gather\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\logits_gather
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_gather
copying deepspeed\inference\v2\kernels\ragged_ops\moe_gather\moe_gather.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_gather
copying deepspeed\inference\v2\kernels\ragged_ops\moe_gather\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_gather
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_scatter
copying deepspeed\inference\v2\kernels\ragged_ops\moe_scatter\moe_scatter.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_scatter
copying deepspeed\inference\v2\kernels\ragged_ops\moe_scatter\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\moe_scatter
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\top_k_gating
copying deepspeed\inference\v2\kernels\ragged_ops\top_k_gating\top_k_gating.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\top_k_gating
copying deepspeed\inference\v2\kernels\ragged_ops\top_k_gating\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\ragged_ops\top_k_gating
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\attn_output_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\embedding_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\invfreq_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\mlp_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\moe_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\norm_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\qkv_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\unembed_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
copying deepspeed\inference\v2\model_implementations\common_parameters\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\common_parameters
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\falcon
copying deepspeed\inference\v2\model_implementations\falcon\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\falcon
copying deepspeed\inference\v2\model_implementations\falcon\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\falcon
copying deepspeed\inference\v2\model_implementations\falcon\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\falcon
copying deepspeed\inference\v2\model_implementations\falcon\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\falcon
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\llama_v2
copying deepspeed\inference\v2\model_implementations\llama_v2\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\llama_v2
copying deepspeed\inference\v2\model_implementations\llama_v2\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\llama_v2
copying deepspeed\inference\v2\model_implementations\llama_v2\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\llama_v2
copying deepspeed\inference\v2\model_implementations\llama_v2\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\llama_v2
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mistral
copying deepspeed\inference\v2\model_implementations\mistral\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mistral
copying deepspeed\inference\v2\model_implementations\mistral\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mistral
copying deepspeed\inference\v2\model_implementations\mistral\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mistral
copying deepspeed\inference\v2\model_implementations\mistral\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mistral
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mixtral
copying deepspeed\inference\v2\model_implementations\mixtral\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mixtral
copying deepspeed\inference\v2\model_implementations\mixtral\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mixtral
copying deepspeed\inference\v2\model_implementations\mixtral\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mixtral
copying deepspeed\inference\v2\model_implementations\mixtral\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\mixtral
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\opt
copying deepspeed\inference\v2\model_implementations\opt\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\opt
copying deepspeed\inference\v2\model_implementations\opt\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\opt
copying deepspeed\inference\v2\model_implementations\opt\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\opt
copying deepspeed\inference\v2\model_implementations\opt\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\opt
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi
copying deepspeed\inference\v2\model_implementations\phi\containers.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi
copying deepspeed\inference\v2\model_implementations\phi\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi
copying deepspeed\inference\v2\model_implementations\phi\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi
copying deepspeed\inference\v2\model_implementations\phi\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi3
copying deepspeed\inference\v2\model_implementations\phi3\containers.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi3
copying deepspeed\inference\v2\model_implementations\phi3\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi3
copying deepspeed\inference\v2\model_implementations\phi3\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi3
copying deepspeed\inference\v2\model_implementations\phi3\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\phi3
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen
copying deepspeed\inference\v2\model_implementations\qwen\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen
copying deepspeed\inference\v2\model_implementations\qwen\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen
copying deepspeed\inference\v2\model_implementations\qwen\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen
copying deepspeed\inference\v2\model_implementations\qwen\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2
copying deepspeed\inference\v2\model_implementations\qwen_v2\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2
copying deepspeed\inference\v2\model_implementations\qwen_v2\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2
copying deepspeed\inference\v2\model_implementations\qwen_v2\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2
copying deepspeed\inference\v2\model_implementations\qwen_v2\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2_moe
copying deepspeed\inference\v2\model_implementations\qwen_v2_moe\container.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2_moe
copying deepspeed\inference\v2\model_implementations\qwen_v2_moe\model.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2_moe
copying deepspeed\inference\v2\model_implementations\qwen_v2_moe\policy.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2_moe
copying deepspeed\inference\v2\model_implementations\qwen_v2_moe\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\qwen_v2_moe
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\attn.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\attn_out.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\embedding.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\mlp.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\qkv.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\types.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\unembed.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
copying deepspeed\inference\v2\model_implementations\sharding\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\model_implementations\sharding
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\attention_configs.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\embedding_config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\linear_config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\moe_config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\norm_config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\unembed_config.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
copying deepspeed\inference\v2\modules\configs\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\configs
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations
copying deepspeed\inference\v2\modules\implementations\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\attention_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\embedding_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\linear_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\moe_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\post_norm_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\pre_norm_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\unembed_base.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
copying deepspeed\inference\v2\modules\interfaces\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\interfaces
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\attention
copying deepspeed\inference\v2\modules\implementations\attention\dense_blocked_attention.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\attention
copying deepspeed\inference\v2\modules\implementations\attention\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\attention
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\embedding
copying deepspeed\inference\v2\modules\implementations\embedding\ragged_embedding.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\embedding
copying deepspeed\inference\v2\modules\implementations\embedding\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\embedding
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\linear
copying deepspeed\inference\v2\modules\implementations\linear\blas_fp_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\linear
copying deepspeed\inference\v2\modules\implementations\linear\quantized_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\linear
copying deepspeed\inference\v2\modules\implementations\linear\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\linear
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\moe
copying deepspeed\inference\v2\modules\implementations\moe\cutlass_multi_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\moe
copying deepspeed\inference\v2\modules\implementations\moe\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\moe
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\post_norm
copying deepspeed\inference\v2\modules\implementations\post_norm\cuda_post_ln.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\post_norm
copying deepspeed\inference\v2\modules\implementations\post_norm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\post_norm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\pre_norm
copying deepspeed\inference\v2\modules\implementations\pre_norm\cuda_pre_ln.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\pre_norm
copying deepspeed\inference\v2\modules\implementations\pre_norm\cuda_pre_rms.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\pre_norm
copying deepspeed\inference\v2\modules\implementations\pre_norm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\pre_norm
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\unembed
copying deepspeed\inference\v2\modules\implementations\unembed\ragged_unembed.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\unembed
copying deepspeed\inference\v2\modules\implementations\unembed\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\modules\implementations\unembed
creating build\lib.win-amd64-cpython-310\deepspeed\model_implementations\diffusers
copying deepspeed\model_implementations\diffusers\unet.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\diffusers
copying deepspeed\model_implementations\diffusers\vae.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\diffusers
copying deepspeed\model_implementations\diffusers\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\diffusers
creating build\lib.win-amd64-cpython-310\deepspeed\model_implementations\features
copying deepspeed\model_implementations\features\cuda_graph.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\features
copying deepspeed\model_implementations\features\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\features
creating build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\clip_encoder.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_base.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_bert.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_bloom.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_gpt.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_llama2.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_megatron_gpt.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_opt.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\ds_transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
copying deepspeed\model_implementations\transformers\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\model_implementations\transformers
creating build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\base.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\base_moe.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\bert.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\bloom.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\clip.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\distil_bert.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\gpt2.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\gptj.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\gptneo.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\gptneox.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\internlm.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\llama.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\llama2.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\megatron_gpt.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\megatron_gpt_moe.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\opt.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\unet.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\vae.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
copying deepspeed\module_inject\containers\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers
creating build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\gated_mlp.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\hybrid_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\hybrid_megatron.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\megatron.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\meta_tensor.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\split_qkv.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
copying deepspeed\module_inject\containers\features\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\module_inject\containers\features
creating build\lib.win-amd64-cpython-310\deepspeed\ops\adagrad
copying deepspeed\ops\adagrad\cpu_adagrad.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adagrad
copying deepspeed\ops\adagrad\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adagrad
creating build\lib.win-amd64-cpython-310\deepspeed\ops\adam
copying deepspeed\ops\adam\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adam
copying deepspeed\ops\adam\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adam
copying deepspeed\ops\adam\multi_tensor_apply.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adam
copying deepspeed\ops\adam\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\adam
creating build\lib.win-amd64-cpython-310\deepspeed\ops\aio
copying deepspeed\ops\aio\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\aio
creating build\lib.win-amd64-cpython-310\deepspeed\ops\compile
copying deepspeed\ops\compile\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\compile
creating build\lib.win-amd64-cpython-310\deepspeed\ops\deepspeed4science
copying deepspeed\ops\deepspeed4science\evoformer_attn.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\deepspeed4science
copying deepspeed\ops\deepspeed4science\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\deepspeed4science
creating build\lib.win-amd64-cpython-310\deepspeed\ops\fp_quantizer
copying deepspeed\ops\fp_quantizer\fp8_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\fp_quantizer
copying deepspeed\ops\fp_quantizer\fp8_gemm_triton.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\fp_quantizer
copying deepspeed\ops\fp_quantizer\quantize.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\fp_quantizer
copying deepspeed\ops\fp_quantizer\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\fp_quantizer
creating build\lib.win-amd64-cpython-310\deepspeed\ops\gds
copying deepspeed\ops\gds\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\gds
creating build\lib.win-amd64-cpython-310\deepspeed\ops\lamb
copying deepspeed\ops\lamb\fused_lamb.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lamb
copying deepspeed\ops\lamb\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lamb
creating build\lib.win-amd64-cpython-310\deepspeed\ops\lion
copying deepspeed\ops\lion\cpu_lion.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lion
copying deepspeed\ops\lion\fused_lion.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lion
copying deepspeed\ops\lion\multi_tensor_apply.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lion
copying deepspeed\ops\lion\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\lion
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\all_ops.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\async_io.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\cpu_adagrad.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\cpu_lion.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\dc.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\evoformer_attn.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\fp_quantizer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\fused_lamb.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\fused_lion.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\gds.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\inference_core_ops.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\inference_cutlass_builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\quantizer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\ragged_ops.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\ragged_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\random_ltd.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\sparse_attn.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\spatial_inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\stochastic_transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\transformer_inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
copying deepspeed\ops\op_builder\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder
creating build\lib.win-amd64-cpython-310\deepspeed\ops\quantizer
copying deepspeed\ops\quantizer\quantizer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\quantizer
copying deepspeed\ops\quantizer\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\quantizer
creating build\lib.win-amd64-cpython-310\deepspeed\ops\random_ltd
copying deepspeed\ops\random_ltd\dropping_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\random_ltd
copying deepspeed\ops\random_ltd\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\random_ltd
creating build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\bert_sparse_self_attention.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\matmul.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\softmax.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\sparse_attention_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\sparse_self_attention.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\sparsity_config.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
copying deepspeed\ops\sparse_attention\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention
creating build\lib.win-amd64-cpython-310\deepspeed\ops\transformer
copying deepspeed\ops\transformer\transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer
copying deepspeed\ops\transformer\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\async_io.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\comm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
copying deepspeed\ops\op_builder\cpu\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\cpu
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\fp_quantizer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\transformer_inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
copying deepspeed\ops\op_builder\hpu\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\hpu
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\cpu_adagrad.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
copying deepspeed\ops\op_builder\mlu\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\mlu
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\async_io.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\cpu_adagrad.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\cpu_lion.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
copying deepspeed\ops\op_builder\npu\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\npu
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
copying deepspeed\ops\op_builder\sdaa\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
copying deepspeed\ops\op_builder\sdaa\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
copying deepspeed\ops\op_builder\sdaa\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
copying deepspeed\ops\op_builder\sdaa\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
copying deepspeed\ops\op_builder\sdaa\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\sdaa
creating build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\async_io.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\builder.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\cpu_adagrad.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\cpu_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\flash_attn.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\fused_adam.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\no_impl.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\packbits.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
copying deepspeed\ops\op_builder\xpu\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\op_builder\xpu
creating build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention\trsrc
copying deepspeed\ops\sparse_attention\trsrc\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention\trsrc
creating build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\bias_add.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\config.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\diffusers_2d_transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\diffusers_attention.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\diffusers_transformer_block.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\ds_attention.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\ds_mlp.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\moe_inference.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\triton_ops.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
copying deepspeed\ops\transformer\inference\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference
creating build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\base.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\bias_add.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\bias_gelu.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\bias_relu.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\bias_residual.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\einsum_sec_sm_ecm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\gated_activation.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\gelu_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\layer_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\linear.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\mlp_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\moe_res_matmul.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\pad_transform.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\pre_rms_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\qkv_gemm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\residual_add.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\rms_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\softmax.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\softmax_context.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\vector_add.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\vector_matmul.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\workspace.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
copying deepspeed\ops\transformer\inference\op_binding\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\op_binding
creating build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\attention.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\gelu.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\layer_norm.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\matmul_ext.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\mlp.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\ops.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\residual_add.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\softmax.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\triton_matmul_kernel.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
copying deepspeed\ops\transformer\inference\triton\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\ops\transformer\inference\triton
creating build\lib.win-amd64-cpython-310\deepspeed\profiling\flops_profiler
copying deepspeed\profiling\flops_profiler\profiler.py -> build\lib.win-amd64-cpython-310\deepspeed\profiling\flops_profiler
copying deepspeed\profiling\flops_profiler\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\profiling\flops_profiler
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\activation_checkpointing
copying deepspeed\runtime\activation_checkpointing\checkpointing.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\activation_checkpointing
copying deepspeed\runtime\activation_checkpointing\config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\activation_checkpointing
copying deepspeed\runtime\activation_checkpointing\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\activation_checkpointing
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\checkpoint_engine
copying deepspeed\runtime\checkpoint_engine\checkpoint_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\checkpoint_engine
copying deepspeed\runtime\checkpoint_engine\nebula_checkpoint_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\checkpoint_engine
copying deepspeed\runtime\checkpoint_engine\torch_checkpoint_engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\checkpoint_engine
copying deepspeed\runtime\checkpoint_engine\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\checkpoint_engine
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\coalesced_collectives.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\compressed.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\hccl.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\mpi.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\nccl.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
copying deepspeed\runtime\comm\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\comm
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\compression
copying deepspeed\runtime\compression\cupy.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\compression
copying deepspeed\runtime\compression\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\compression
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline
copying deepspeed\runtime\data_pipeline\config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline
copying deepspeed\runtime\data_pipeline\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline
copying deepspeed\runtime\data_pipeline\curriculum_scheduler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline
copying deepspeed\runtime\data_pipeline\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\domino
copying deepspeed\runtime\domino\async_linear.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\domino
copying deepspeed\runtime\domino\transformer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\domino
copying deepspeed\runtime\domino\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\domino
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16
copying deepspeed\runtime\fp16\fused_optimizer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16
copying deepspeed\runtime\fp16\loss_scaler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16
copying deepspeed\runtime\fp16\unfused_optimizer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16
copying deepspeed\runtime\fp16\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\engine.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\module.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\p2p.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\schedule.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\topology.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
copying deepspeed\runtime\pipe\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\pipe
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\aio_config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\async_swapper.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\constants.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\optimizer_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\partitioned_optimizer_swapper.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\partitioned_param_swapper.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\pipelined_optimizer_swapper.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
copying deepspeed\runtime\swap_tensor\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\swap_tensor
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\tensor_parallel
copying deepspeed\runtime\tensor_parallel\config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\tensor_parallel
copying deepspeed\runtime\tensor_parallel\tp_manager.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\tensor_parallel
copying deepspeed\runtime\tensor_parallel\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\tensor_parallel
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\contiguous_memory_allocator.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\linear.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\mics.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\mics_utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\offload_config.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\offload_states.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\parameter_offload.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\partitioned_param_coordinator.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\partitioned_param_profiler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\partition_parameters.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\stage3.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\stage_1_and_2.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\test.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\tiling.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
copying deepspeed\runtime\zero\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\zero
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
copying deepspeed\runtime\data_pipeline\data_routing\basic_layer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
copying deepspeed\runtime\data_pipeline\data_routing\helper.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
copying deepspeed\runtime\data_pipeline\data_routing\scheduler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
copying deepspeed\runtime\data_pipeline\data_routing\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
copying deepspeed\runtime\data_pipeline\data_routing\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_routing
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\data_analyzer.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\data_sampler.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\indexed_dataset.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\utils.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\variable_batch_size_and_lr.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
copying deepspeed\runtime\data_pipeline\data_sampling\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\data_pipeline\data_sampling
creating build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16\onebit
copying deepspeed\runtime\fp16\onebit\adam.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16\onebit
copying deepspeed\runtime\fp16\onebit\lamb.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16\onebit
copying deepspeed\runtime\fp16\onebit\zoadam.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16\onebit
copying deepspeed\runtime\fp16\onebit\__init__.py -> build\lib.win-amd64-cpython-310\deepspeed\runtime\fp16\onebit
running egg_info
writing deepspeed.egg-info\PKG-INFO
writing dependency_links to deepspeed.egg-info\dependency_links.txt
writing entry points to deepspeed.egg-info\entry_points.txt
writing requirements to deepspeed.egg-info\requires.txt
writing top-level names to deepspeed.egg-info\top_level.txt
reading manifest file 'deepspeed.egg-info\SOURCES.txt'
reading manifest template 'MANIFEST_win.in'
warning: no previously-included files matching '*.cpp' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.h' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cu' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cuh' found under directory 'deepspeed\ops\csrc'
warning: no previously-included files matching '*.cc' found under directory 'deepspeed\ops\csrc'
no previously-included directories found matching 'op_builder'
no previously-included directories found matching 'accelerator'
adding license file 'LICENSE'
writing manifest file 'deepspeed.egg-info\SOURCES.txt'
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops.bias_activations' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops.bias_activations' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops.bias_activations' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops.bias_activations' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops.bias_activations' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops.cuda_linear' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops.cuda_linear' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops.cuda_linear' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops.cuda_linear' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops.cuda_linear' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.kernels.core_ops.gated_activations' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.kernels.core_ops.gated_activations' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.kernels.core_ops.gated_activations' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.kernels.core_ops.gated_activations' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.kernels.core_ops.gated_activations' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\setuptools\command\build_py.py:212: _Warning: Package 'deepspeed.inference.v2.ragged.csrc' is absent from the `packages` configuration.
!!

        ********************************************************************************
        ############################
        # Package would be ignored #
        ############################
        Python recognizes 'deepspeed.inference.v2.ragged.csrc' as an importable package[^1],
        but it is absent from setuptools' `packages` configuration.

        This leads to an ambiguous overall configuration. If you want to distribute this
        package, please make sure that 'deepspeed.inference.v2.ragged.csrc' is explicitly added
        to the `packages` configuration field.

        Alternatively, you can also rely on setuptools' discovery methods
        (for example by using `find_namespace_packages(...)`/`find_namespace:`
        instead of `find_packages(...)`/`find:`).

        You can read more about "package discovery" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/package_discovery.html

        If you don't want 'deepspeed.inference.v2.ragged.csrc' to be distributed and are
        already explicitly excluding 'deepspeed.inference.v2.ragged.csrc' via
        `find_namespace_packages(...)/find_namespace` or `find_packages(...)/find`,
        you can try to use `exclude_package_data`, or `include-package-data=False` in
        combination with a more fine grained `package-data` configuration.

        You can read more about "package data files" on setuptools documentation page:

        - https://setuptools.pypa.io/en/latest/userguide/datafiles.html


        [^1]: For Python, any directory (with suitable naming) can be imported,
              even if it does not contain any `.py` files.
              On the other hand, currently there is no concept of package data
              directory, all directories are treated like packages.
        ********************************************************************************

!!
  check.warn(importable)
copying deepspeed\inference\v2\kernels\core_ops\core_ops.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\bias_activation.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\bias_activation_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\layer_norm.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\layer_norm_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\linear_kernels.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\linear_kernels_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\gated_activation_kernels.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\gated_activation_kernels_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
creating build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged\csrc
copying deepspeed\inference\v2\ragged\csrc\fast_host_buffer.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged\csrc
copying deepspeed\inference\v2\ragged\csrc\ragged_ops.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged\csrc
copying deepspeed\inference\v2\ragged\csrc\fast_host_buffer.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged\csrc
copying deepspeed\inference\v2\ragged\csrc\ragged_ops.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\ragged\csrc
copying deepspeed\inference\v2\kernels\core_ops\core_ops.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\bias_activation.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\bias_activations\bias_activation_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\bias_activations
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\layer_norm.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm\layer_norm_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_layer_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\linear_kernels.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_linear\linear_kernels_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_linear
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm\rms_norm_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\cuda_rms_norm
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\gated_activation_kernels.cpp -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
copying deepspeed\inference\v2\kernels\core_ops\gated_activations\gated_activation_kernels_cuda.cu -> build\lib.win-amd64-cpython-310\deepspeed\inference\v2\kernels\core_ops\gated_activations
copying deepspeed\ops\sparse_attention\trsrc\matmul.tr -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention\trsrc
copying deepspeed\ops\sparse_attention\trsrc\softmax_bwd.tr -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention\trsrc
copying deepspeed\ops\sparse_attention\trsrc\softmax_fwd.tr -> build\lib.win-amd64-cpython-310\deepspeed\ops\sparse_attention\trsrc
running build_ext
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\utils\cpp_extension.py:382: UserWarning: Error checking compiler version for cl: [WinError 2] The system cannot find the file specified
  warnings.warn(f'Error checking compiler version for {compiler}: {error}')
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\utils\cpp_extension.py:416: UserWarning: The detected CUDA version (12.6) has a minor version mismatch with the version that was used to compile PyTorch (12.1). Most likely this shouldn't be a problem.
  warnings.warn(CUDA_MISMATCH_WARN.format(cuda_str_version, torch.version.cuda))
building 'deepspeed.ops.adam.fused_adam_op' extension
creating build\temp.win-amd64-cpython-310\Release\csrc\adam
cl.exe /c /nologo /O2 /W3 /GL /DNDEBUG /MD -IC:\AI\ACE-Step\DeepSpeed\csrc\includes -IC:\AI\ACE-Step\DeepSpeed\csrc\adam -IC:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\include -IC:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\include\torch\csrc\api\include -IC:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\include\TH -IC:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.6\include" -IC:\Users\DjZaN\miniconda3\envs\ace_step\include -IC:\Users\DjZaN\miniconda3\envs\ace_step\Include /EHsc /Tpcsrc/adam/fused_adam_frontend.cpp /Fobuild\temp.win-amd64-cpython-310\Release\csrc\adam\fused_adam_frontend.obj /MD /wd4819 /wd4251 /wd4244 /wd4267 /wd4275 /wd4018 /wd4190 /wd4624 /wd4067 /wd4068 /EHsc -O2 -DVERSION_GE_1_1 -DVERSION_GE_1_3 -DVERSION_GE_1_5 -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=fused_adam_op -D_GLIBCXX_USE_CXX11_ABI=0 /std:c++17
error: command 'cl.exe' failed: None

ERROR Backend subprocess exited when trying to invoke build_wheel
(ace_step) C:\AI\ACE-Step\DeepSpeed>

ASSISTANT

The key error in your DeepSpeed build logs is:

```
C:\Users\DjZaN\miniconda3\envs\ace_step\lib\site-packages\torch\utils\cpp_extension.py:382: UserWarning: Error checking compiler version for cl: [WinError 2] The system cannot find the file specified
...
error: command 'cl.exe' failed: None
```

This means the Windows C++ compiler (`cl.exe`) was **not found** on your PATH, hence the build failed.

---

# How to fix `cl.exe` not found and build DeepSpeed on Windows

### 1. Install Visual Studio Build Tools (if not done yet)

- Download and install **Visual Studio Build Tools** from here:

https://visualstudio.microsoft.com/visual-cpp-build-tools/

- During installation, **select the "Desktop development with C++" workload**.
- Also make sure to install the Windows 10 SDK or Windows 11 SDK depending on your OS.

---

### 2. Use the "Developer Command Prompt for Visual Studio"

- After installation, open the **"x64 Native Tools Command Prompt for VS 20xx"** from your Start menu.

This sets up environment variables so `cl.exe` is on your path.

- Then activate your conda environment inside this prompt:

```bash
conda activate ace_step
```

- Navigate to your DeepSpeed folder and run your build batch/script:

```bash
cd C:\AI\ACE-Step\DeepSpeed
build_win.bat
```

> **Important:** Running inside this Developer Command Prompt ensures `cl.exe` (MSVC compiler) is accessible.

---

### 3. Verify `cl.exe` is found

Before running the build, you can check if `cl.exe` is available by typing:

```bash
cl
```

You should see Microsoft (R) C/C++ Optimizing Compiler info instead of a "not recognized" error.

---

### 4. CUDA version mismatch warnings

You're seeing these warnings:

```
Installed CUDA version 12.6 does not match the version torch was compiled with 12.1 but since the APIs are compatible, accepting this combination
```

This warning is usually safe to ignore if you're sure your CUDA driver and toolkit are installed correctly.

If you want to be 100% sure: you can install PyTorch built against CUDA 12.6 (if available) or downgrade your local CUDA toolkit to 12.1.

---

### 5. Additional notes

- Ensure your PyTorch and CUDA toolkit versions are compatible.
- Make sure your Python version and compiler are compatible (your build is on Python 3.10 which is fine).
- You can also try to install `ninja` beforehand, which speeds up and sometimes fixes builds:

```bash
pip install ninja
```

---

# Summary of steps

1. Install Visual Studio Build Tools with C++ workload.
2. Open **x64 Native Tools Command Prompt for VS**.
3. Activate conda: `conda activate ace_step`.
4. Run DeepSpeed build script: `build_win.bat`.
5. Verify successful build and check for errors.

---

If you still get errors, please share them, and I can help further.

---

# Optional: Use WSL2 instead

If this proves complicated or errors persist, the easiest and most stable way to use DeepSpeed on Windows is:

- Install WSL2 (Ubuntu Linux on Windows).
- Install CUDA for WSL if using GPU.
- Install DeepSpeed following Linux instructions inside WSL.

This environment is far more tested and supported for DeepSpeed.

---

Let me know if you want guidance setting up WSL2 + DeepSpeed too!

turns-00078.parquet:16172

7241f414802dcf7592432647

turn 4/6gpt-4.1-mini-2025-04-14EnglishUnited States521 words

degenerate_repetitionAbsentFinal dense release

USER

(ace_step) C:\AI\ACE-Step\DeepSpeed>build_win.bat
* Getting build dependencies for wheel...
OMP: Error #15: Initializing libiomp5md.dll, but found libiomp5md.dll already initialized.
OMP: Hint This means that multiple copies of the OpenMP runtime have been linked into the program. That is dangerous, since it can degrade performance or cause incorrect results. The best thing to do is to ensure that only a single OpenMP runtime is linked into the process, e.g. by avoiding static linking of the OpenMP runtime in any library. As an unsafe, unsupported, undocumented workaround you can set the environment variable KMP_DUPLICATE_LIB_OK=TRUE to allow the program to continue to execute, but that may cause crashes or silently produce incorrect results. For more information, please see http://www.intel.com/software/products/support/.

ERROR Backend subprocess exited when trying to invoke get_requires_for_build_wheel

ASSISTANT

The error message:

```
OMP: Error #15: Initializing libiomp5md.dll, but found libiomp5md.dll already initialized.
OMP: Hint This means that multiple copies of the OpenMP runtime have been linked into the program. ...
```

indicates that **multiple conflicting OpenMP runtimes (libiomp5md.dll)** are loaded at the same time during the build process. This is a known issue on Windows, especially when different Python packages or libraries link their own OpenMP runtimes.

---

## What causes this?

- Intel OpenMP (`libiomp5md.dll`) is bundled with several Python packages (like `numpy`, `scipy`, `torch`, `scikit-learn`) and your toolchain.
- If different versions or static/dynamic linking mix happens, this conflict arises.
- This error often pops up during compilation or running code that uses OpenMP in Python extensions.

---

## How to fix it?

### Short-term workaround (unsafe)

Set the environment variable `KMP_DUPLICATE_LIB_OK=TRUE` **before** running the build command to ignore the error (not recommended for production, but OK for building):

In your command prompt (Windows):

```cmd
set KMP_DUPLICATE_LIB_OK=TRUE
build_win.bat
```

Or, if you prefer a one-liner PowerShell:

```powershell
$env:KMP_DUPLICATE_LIB_OK="TRUE"; .\build_win.bat
```

This lets the build continue but may cause instability in some cases.

---

### Better approach: try these steps

1. **Use a "clean" environment to build**, minimize packages that load `libiomp5md.dll` before your build (especially libraries that import numpy/scipy/sklearn before the build).

2. **Check your PATH and environment** for conflicting Intel OpenMP DLLs:
- Sometimes Anaconda/miniconda `Library\bin` folder ships a `libiomp5md.dll`.
- Other software might also add conflicting DLLs.

3. **Ensure consistent tooling**:
- Use a fresh conda environment dedicated to building DeepSpeed.
- Install only minimal necessary packages (e.g. `python`, `pip`, `wheel`, `setuptools`, `torch`).

4. You can explicitly **unset or avoid early import** of packages that load OpenMP before running the build.

---

### Example: Try in a fresh cmd session:

```cmd
conda create -n ds_test python=3.10
conda activate ds_test
pip install torch
set KMP_DUPLICATE_LIB_OK=TRUE
git clone https://github.com/microsoft/DeepSpeed.git
cd DeepSpeed
build_win.bat
```

This minimizes conflicting DLLs loaded.

---

## Additional resources:

- Intel OpenMP error explanation: https://software.intel.com/content/www/us/en/develop/articles/openmp-run-time-warnings-and-errors.html
- GitHub issues opened about this happen in PyTorch and DeepSpeed builds as well.

---

## TL;DR

- Set `KMP_DUPLICATE_LIB_OK=TRUE` environment variable before build to bypass error.
- Consider building in a clean conda env with minimal packages.
- Avoid importing numpy/sklearn or other complex libs before build.

---

If you want, I can help you script a clean build environment or troubleshoot more in-depth!