Spaces:

Fabrice-TIERCELIN

/

HunyuanVideo

like 11

Running

Hi,

I have changed the path to ckpts. You can retry in 3 ways:

• Synchronize your space from this one
• Replace tencent_HunyuanVideo by ckpts in app.py
• Or duplicate your space a second time

I duplicated the Space again and got this error:
ValueError: Invalid model path: ckpts/hunyuan-video-t2v-720p/transformers/mp_rank_00_model_states.pt

So the same error.

I have added some logs. Do you see in your logs those ones?
initialize_model: ...
models_root exists: ...
Model initialized: ...

And also this one and the following?
What is dit_weight: ckpts/hunyuan-video-t2v-720p/transformers/mp_rank_00_model_states.pt

PS: I have slightly changed the code, that may fix the space

This is the output when I just tried to duplicate. It is different from the previous errors.

runtime error
Exit code: 1. Reason: A

mp_rank_00_model_states_fp8.pt: 90%|████████▉ | 11.9G/13.2G [00:09<00:01, 1.31GB/s]
mp_rank_00_model_states_fp8.pt: 100%|█████████▉| 13.2G/13.2G [00:10<00:00, 1.30GB/s]

mp_rank_00_model_states_fp8_map.pt: 0%| | 0.00/104k [00:00<?, ?B/s]
mp_rank_00_model_states_fp8_map.pt: 100%|██████████| 104k/104k [00:00<00:00, 39.7MB/s]

hunyuan-video-t2v-720p/vae/config.json: 0%| | 0.00/785 [00:00<?, ?B/s]
hunyuan-video-t2v-720p/vae/config.json: 100%|██████████| 785/785 [00:00<00:00, 8.40MB/s]

pytorch_model.pt: 0%| | 0.00/986M [00:00<?, ?B/s]

pytorch_model.pt: 100%|█████████▉| 986M/986M [00:01<00:00, 918MB/s]
pytorch_model.pt: 100%|█████████▉| 986M/986M [00:02<00:00, 460MB/s]
initialize_model: ckpts
models_root exists: ckpts
2025-01-03 07:23:31.750 | INFO | hyvideo.inference:from_pretrained:154 - Got text-to-video model root path: ckpts
2025-01-03 07:23:31.974 | INFO | hyvideo.inference:from_pretrained:189 - Building model...
What is dit_weight: ckpts/hunyuan-video-t2v-720p/transformers/mp_rank_00_model_states.pt
dit_weight.exists(): False
dit_weight.is_file(): False
dit_weight.is_dir(): False
dit_weight.is_symlink(): False
Traceback (most recent call last):
File "/home/user/app/app.py", line 170, in
demo = create_demo("ckpts")
File "/home/user/app/app.py", line 94, in create_demo
model = initialize_model(model_path)
File "/home/user/app/app.py", line 40, in initialize_model
hunyuan_video_sampler = HunyuanVideoSampler.from_pretrained(models_root_path, args=args)
File "/home/user/app/hyvideo/inference.py", line 203, in from_pretrained
model = Inference.load_state_dict(args, model, pretrained_model_path)
File "/home/user/app/hyvideo/inference.py", line 314, in load_state_dict
print('dit_weight.is_junction(): ' + str(dit_weight.is_junction()))
AttributeError: 'PosixPath' object has no attribute 'is_junction'
Container logs:

===== Application Startup at 2025-01-03 06:20:03 =====

The cache for model files in Transformers v4.22.0 has been updated. Migrating your old cache. This is a one-time only operation. You can interrupt this and resume the migration later on by calling transformers.utils.move_cache().

initialize_model: ckpts
models_root exists: ckpts
2025-01-03 07:23:31.750 | INFO | hyvideo.inference:from_pretrained:154 - Got text-to-video model root path: ckpts
2025-01-03 07:23:31.974 | INFO | hyvideo.inference:from_pretrained:189 - Building model...
What is dit_weight: ckpts/hunyuan-video-t2v-720p/transformers/mp_rank_00_model_states.pt
dit_weight.exists(): False
dit_weight.is_file(): False
dit_weight.is_dir(): False
dit_weight.is_symlink(): False
Traceback (most recent call last):
File "/home/user/app/app.py", line 170, in
demo = create_demo("ckpts")
File "/home/user/app/app.py", line 94, in create_demo
model = initialize_model(model_path)
File "/home/user/app/app.py", line 40, in initialize_model
hunyuan_video_sampler = HunyuanVideoSampler.from_pretrained(models_root_path, args=args)
File "/home/user/app/hyvideo/inference.py", line 203, in from_pretrained
model = Inference.load_state_dict(args, model, pretrained_model_path)
File "/home/user/app/hyvideo/inference.py", line 314, in load_state_dict
print('dit_weight.is_junction(): ' + str(dit_weight.is_junction()))
AttributeError: 'PosixPath' object has no attribute 'is_junction'
The cache for model files in Transformers v4.22.0 has been updated. Migrating your old cache. This is a one-time only operation. You can interrupt this and resume the migration later on by calling transformers.utils.move_cache().

OK, you can retry. (now it download with snapshot and not file by file)

May I guess it's working now? 🙂

No, still getting and error. I just kind of got frustrated and gave up.

runtime error
Exit code: 1. Reason: coder model (llm) from: ./ckpts/text_encoder
Traceback (most recent call last):
File "/home/user/.pyenv/versions/3.10.16/lib/python3.10/site-packages/transformers/utils/hub.py", line 403, in cached_file
resolved_file = hf_hub_download(
File "/home/user/.pyenv/versions/3.10.16/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 106, in _inner_fn
validate_repo_id(arg_value)
File "/home/user/.pyenv/versions/3.10.16/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 154, in validate_repo_id
raise HFValidationError(
huggingface_hub.errors.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': './ckpts/text_encoder'. Use repo_type argument if needed.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "/home/user/app/app.py", line 167, in
demo = create_demo("ckpts")
File "/home/user/app/app.py", line 86, in create_demo
model = initialize_model(model_path)
File "/home/user/app/app.py", line 32, in initialize_model
hunyuan_video_sampler = HunyuanVideoSampler.from_pretrained(models_root_path, args=args)
File "/home/user/app/hyvideo/inference.py", line 241, in from_pretrained
text_encoder = TextEncoder(
File "/home/user/app/hyvideo/text_encoder/init.py", line 180, in init
self.model, self.model_path = load_text_encoder(
File "/home/user/app/hyvideo/text_encoder/init.py", line 36, in load_text_encoder
text_encoder = AutoModel.from_pretrained(
File "/home/user/.pyenv/versions/3.10.16/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 487, in from_pretrained
resolved_config_file = cached_file(
File "/home/user/.pyenv/versions/3.10.16/lib/python3.10/site-packages/transformers/utils/hub.py", line 469, in cached_file
raise EnvironmentError(
OSError: Incorrect path_or_model_id: './ckpts/text_encoder'. Please provide either the path to a local folder or the repo_id of a model on the Hub.

FYI, I have updated the space thanks to your log

@Fabrice-TIERCELIN

I duplicated the space and selected ZeroGPU.

Got this error

Traceback (most recent call last):
File "/home/user/app/app.py", line 22, in
preprocess_text_encoder_tokenizer(input_dir = "ckpts/llava-llama-3-8b-v1_1-transformers", output_dir = "ckpts/text_encoder")
TypeError: preprocess_text_encoder_tokenizer() got an unexpected keyword argument 'input_dir'

Hi,

I have fixed this bug. You can retry in 3 ways:

• Synchronize your space from this one (click on Settings and then on Synchronize)
• Copy/paste the new code in app.py
• Or duplicate your space a second time

@Fabrice-TIERCELIN

Thanks. But after clicking "Generate" , I got this error,

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/gradio/queueing.py", line 536, in process_events
response = await route_utils.call_process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1935, in process_api
result = await self.call_function(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1520, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "/usr/local/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2505, in run_sync_in_worker_thread
return await future
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 1005, in run
result = context.run(func, *args)
File "/usr/local/lib/python3.10/site-packages/gradio/utils.py", line 826, in wrapper
response = f(*args, **kwargs)
File "/home/user/app/app.py", line 161, in
fn=lambda *inputs: generate_video(model, *inputs),
File "/usr/local/lib/python3.10/site-packages/spaces/zero/wrappers.py", line 201, in gradio_handler
worker.arg_queue.put(((args, kwargs), GradioPartialContext.get()))
File "/usr/local/lib/python3.10/site-packages/spaces/utils.py", line 54, in put
raise PicklingError(message)
_pickle.PicklingError: cannot pickle '_io.TextIOWrapper' object

I have added the dill library because some people has solved the problem this way. You can retry.

@Fabrice-TIERCELIN sorry. Still same error persists. I synced the changes

I have changed a syntax that is possibly the root cause. I have also added some logs to know the moment of the error. Do you see the following messages in the logs?

1. generate_video (prompt:
2. generate_video_gpu (prompt:
3. Predicting video...
4. Video predicted

@Fabrice-TIERCELIN

I got this . Right from the beginning.

Running on local URL: http://0.0.0.0:7860

To create a public link, set share=True in launch().
generate_video (prompt: A cat walks on the grass, realistic style.)
Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/spaces/utils.py", line 46, in put
super().put(obj)
File "/usr/local/lib/python3.10/multiprocessing/queues.py", line 371, in put
obj = _ForkingPickler.dumps(obj)
File "/usr/local/lib/python3.10/multiprocessing/reduction.py", line 51, in dumps
cls(buf, protocol).dump(obj)
TypeError: cannot pickle '_io.TextIOWrapper' object

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/gradio/queueing.py", line 536, in process_events
response = await route_utils.call_process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1935, in process_api
result = await self.call_function(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1520, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "/usr/local/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2505, in run_sync_in_worker_thread
return await future
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 1005, in run
result = context.run(func, *args)
File "/usr/local/lib/python3.10/site-packages/gradio/utils.py", line 826, in wrapper
response = f(*args, **kwargs)
File "/home/user/app/app.py", line 61, in generate_video
return generate_video_gpu(
File "/usr/local/lib/python3.10/site-packages/spaces/zero/wrappers.py", line 201, in gradio_handler
worker.arg_queue.put(((args, kwargs), GradioPartialContext.get()))
File "/usr/local/lib/python3.10/site-packages/spaces/utils.py", line 54, in put
raise PicklingError(message)
_pickle.PicklingError: cannot pickle '_io.TextIOWrapper' object

Hi @hysts & @Mar2Ding ,

You said here that you have solved the same error that @RageshAntony and me are facing in the comment above when using this space. How do you solved it? I can't find the fix in the two PR diffs.

@RageshAntony is running the current duplicated space on ZERO and me has no GPU.

I was able to duplicate with an A100, but got this error when trying to generate:

File "/home/user/app/hyvideo/modules/models.py", line 204, in forward
attn = attention(
File "/home/user/app/hyvideo/modules/attenion.py", line 108, in attention
x = flash_attn_varlen_func(
TypeError: 'NoneType' object is not callable

I have set torch instead of flash.

@johnblues , you can retry on A100.

@Fabrice-TIERCELIN
After synced the build didn't start and it threw a configuration error :

No candidate PyTorch version found for ZeroGPU

I have removed torch==2.5.1 in the requirements as it has solved the problem in a similar case.

@RageshAntony , you can retry.

Having an issue uploading the video. Here is a screenshot.

So it's not working yet.

Try a Guidance Scale to 7

And if it's not better, try to increase the steps as much as you can for max 2 minutes of generation.

Still the same.

OK, if you want, you can try with any other parameters. Change the prompt to simple one like "dog", "tree".

Meanwhile, I will update the code and test options. Do not synchronize now.

Are you still on ZeroGPU or dedicated GPU?

Now you can synchronize.

@Fabrice-TIERCELIN

sorry to say. Still the same issue.

Running on local URL: http://0.0.0.0:7860

To create a public link, set share=True in launch().
generate_video (prompt: A cat walks on the grass, steampunk world style )
Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/spaces/utils.py", line 46, in put
super().put(obj)
File "/usr/local/lib/python3.10/multiprocessing/queues.py", line 371, in put
obj = _ForkingPickler.dumps(obj)
File "/usr/local/lib/python3.10/multiprocessing/reduction.py", line 51, in dumps
cls(buf, protocol).dump(obj)
TypeError: cannot pickle '_io.TextIOWrapper' object

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/gradio/queueing.py", line 536, in process_events
response = await route_utils.call_process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1935, in process_api
result = await self.call_function(
File "/usr/local/lib/python3.10/site-packages/gradio/blocks.py", line 1520, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "/usr/local/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2461, in run_sync_in_worker_thread
return await future
File "/usr/local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 962, in run
result = context.run(func, *args)
File "/usr/local/lib/python3.10/site-packages/gradio/utils.py", line 826, in wrapper
response = f(*args, **kwargs)
File "/home/user/app/app.py", line 61, in generate_video
return generate_video_gpu(
File "/usr/local/lib/python3.10/site-packages/spaces/zero/wrappers.py", line 201, in gradio_handler
worker.arg_queue.put(((args, kwargs), GradioPartialContext.get()))
File "/usr/local/lib/python3.10/site-packages/spaces/utils.py", line 54, in put
raise PicklingError(message)
_pickle.PicklingError: cannot pickle '_io.TextIOWrapper' object

@Fabrice-TIERCELIN

my duplicated repo .
https://huggingface.co/spaces/RageshAntony/HunyuanVideo

I still just get noise output on the A100. If I use a Zero Space, I get an error.

OK, I have replaced all the single quotes by double quotes to get rid off the error _pickle.PicklingError: cannot pickle '_io.TextIOWrapper' object

Can someone synchronize with a ZERO and tell me if we get rid off the error _pickle.PicklingError: cannot pickle '_io.TextIOWrapper' object?

It would be a step forward and if it works, I will raise this resolution to the project.