Emotion-LLaMA
Emotion-LLaMA copied to clipboard
ZebangCheng
部署本地demo时遇到这个问题，是否和没有video文件夹有关？
/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/timm/models/hub.py:4: FutureWarning: Importing from timm.models.hub is deprecated, please import via timm.models
warnings.warn(f"Importing from {name} is deprecated, please import via timm.models", FutureWarning)
/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/timm/models/layers/init.py:48: FutureWarning: Importing from timm.models.layers is deprecated, please import via timm.layers
warnings.warn(f"Importing from {name} is deprecated, please import via timm.layers", FutureWarning)
/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/timm/models/registry.py:4: FutureWarning: Importing from timm.models.registry is deprecated, please import via timm.models
warnings.warn(f"Importing from {name} is deprecated, please import via timm.models", FutureWarning)
Initializing Chat
model_config: {'arch': 'minigpt_v2', 'image_size': 448, 'drop_path_rate': 0, 'use_grad_checkpoint': False, 'vit_precision': 'fp16', 'freeze_vit': True, 'prompt': '', 'llama_model': '/home/marsdy/Emotion-LLaMA/checkpoints/Llama-2-7b-chat-hf', 'lora_r': 64, 'lora_alpha': 16, 'model_type': 'pretrain', 'max_txt_len': 500, 'end_sym': '', 'low_resource': True, 'prompt_template': '[INST] {} [/INST]', 'ckpt': '/home/marsdy/Emotion-LLaMA/checkpoints/save_checkpoint/Emoation_LLaMA.pth'}
The load_in_4bit and load_in_8bit arguments are deprecated and will be removed in the future versions. Please, pass a BitsAndBytesConfig object in quantization_config argument instead.
Loading checkpoint shards: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:36<00:00, 18.29s/it]
WARNING:py.warnings:/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/peft/utils/other.py:145: FutureWarning: prepare_model_for_int8_training is deprecated and will be removed in a future version. Use prepare_model_for_kbit_training instead.
warnings.warn(
loraconfig: LoraConfig(peft_type=<PeftType.LORA: 'LORA'>, auto_mapping=None, base_model_name_or_path=None, revision=None, task_type='CAUSAL_LM', inference_mode=False, r=64, target_modules={'v_proj', 'q_proj'}, lora_alpha=16, lora_dropout=0.05, fan_in_fan_out=False, bias='none', use_rslora=False, modules_to_save=None, init_lora_weights=True, layers_to_transform=None, layers_pattern=None, rank_pattern={}, alpha_pattern={}, megatron_config=None, megatron_core='megatron.core', loftq_config={}, use_dora=False)
trainable params: 33,554,432 || all params: 6,771,970,048 || trainable%: 0.49548996469513035
Position interpolate from 16x16 to 32x32
Load Minigpt-4-LLM Checkpoint: /home/marsdy/Emotion-LLaMA/checkpoints/save_checkpoint/Emoation_LLaMA.pth
Traceback (most recent call last):
File "/home/marsdy/Emotion-LLaMA/app.py", line 61, in load_in_4bit and load_in_8bit arguments are deprecated and will be removed in the future versions. Please, pass a BitsAndBytesConfig object in quantization_config argument instead.
Loading checkpoint shards: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:25<00:00, 12.64s/it]
WARNING:py.warnings:/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/peft/utils/other.py:145: FutureWarning: prepare_model_for_int8_training is deprecated and will be removed in a future version. Use prepare_model_for_kbit_training instead.
warnings.warn(
loraconfig: LoraConfig(peft_type=<PeftType.LORA: 'LORA'>, auto_mapping=None, base_model_name_or_path=None, revision=None, task_type='CAUSAL_LM', inference_mode=False, r=64, target_modules={'v_proj', 'q_proj'}, lora_alpha=16, lora_dropout=0.05, fan_in_fan_out=False, bias='none', use_rslora=False, modules_to_save=None, init_lora_weights=True, layers_to_transform=None, layers_pattern=None, rank_pattern={}, alpha_pattern={}, megatron_config=None, megatron_core='megatron.core', loftq_config={}, use_dora=False) trainable params: 33,554,432 || all params: 6,771,970,048 || trainable%: 0.49548996469513035 Position interpolate from 16x16 to 32x32 Load Minigpt-4-LLM Checkpoint: /home/marsdy/Emotion-LLaMA/checkpoints/save_checkpoint/Emoation_LLaMA.pth WARNING:py.warnings:/home/marsdy/anaconda3/envs/llama/lib/python3.9/site-packages/gradio/layouts/column.py:55: UserWarning: 'scale' value should be an integer. Using 0.5 will cause issues. warnings.warn(
Traceback (most recent call last):
File "/home/marsdy/Emotion-LLaMA/app.py", line 635, in
第一个问题是显存不够：torch.cuda.OutOfMemoryError: CUDA out of memory. 第二个问题是gradio版本不对，如果你的gradio版本高了，gr.Examples组件就会报错。建议使用gradio 3.47.1版本。
(llama) [u202320085410023@gpu2 Emotion-LLaMA]$ python app.py
===================================BUG REPORT=================================== Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues
Initializing Chat model_config: {'arch': 'minigpt_v2', 'image_size': 448, 'drop_path_rate': 0, 'use_grad_checkpoint': False, 'vit_precision': 'fp16', 'freeze_vit': True, 'prompt': '', 'llama_model': '/hdd/u202320085410023/Emotion-LLaMA/checkpoints/Llama-2-7b-chat-hf', 'lora_r': 64, 'lora_alpha': 16, 'model_type': 'pretrain', 'max_txt_len': 500, 'end_sym': '', 'low_resource': True, 'prompt_template': '[INST] {} [/INST]', 'ckpt': '/hdd/u202320085410023/Emotion-LLaMA/checkpoints/save_checkpoint/Emoation_LLaMA.pth'} Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████| 2/2 [00:13<00:00, 6.91s/it] loraconfig: LoraConfig(peft_type=<PeftType.LORA: 'LORA'>, base_model_name_or_path=None, task_type='CAUSAL_LM', inference_mode=False, r=64, target_modules=['q_proj', 'v_proj'], lora_alpha=16, lora_dropout=0.05, merge_weights=False, fan_in_fan_out=False, enable_lora=None, bias='none', modules_to_save=None) trainable params: 33554432 || all params: 6771970048 || trainable%: 0.49548996469513035 54%|██████████████████████████████████████████████▌ | 1.02G/1.89G [02:00<01:25, 10.8MB/s] 您好，我在本地部署demo时总是遇到上面这种情况，没有报错，但是进度进行一部分就卡住不再进行了，过一段时间后还是如此。重新试了很多次，每次卡住的进度都不同，请问这是什么原因，怎么解决呢？
1.02G/1.89G [02:00<01:25, 10.8MB/s] 这一部分应该是从hugging face下载模型（BERT或者EVA视觉Encoder）卡住了。尝试使用hugging face镜像网站试试。
第一个问题是显存不够：torch.cuda.OutOfMemoryError: CUDA out of memory. 第二个问题是gradio版本不对，如果你的gradio版本高了，gr.Examples组件就会报错。建议使用gradio 3.47.1版本。
我换成3.47.1版本会报如下错误：PermissionError: [Errno 13] Permission denied: '/tmp/gradio/bc589d81dda901fbca764ae4de637e1ae0ebdf08' IMPORTANT: You are using gradio version 3.47.1, however version 4.44.1 is available, please upgrade.
抱歉，前段时间在忙，没有及时回复。
你这个问题和版本没有关系。IMPORTANT: You are using gradio version 3.47.1, however version 4.44.1 is available, please upgrade.这是警告，不是代码错误的核心原因。
真正的错误原因是PermissionError。因为demo需要提取视频中的audio，并将audio临时保持到本地，具体代码如下：
def extract_audio_from_video(video_path):
video = VideoFileClip(video_path)
audio = video.audio
# audio.write_audiofile("audio.wav")
audio_path = "audio.wav"
audio.write_audiofile(audio_path, fps=16000, codec='pcm_s16le', ffmpeg_params=['-ac', '1'])
samples, sr = sf.read(audio_path)
return samples, sr
所以你需要基于代码或者用户写入的权限。
我在网络上检索后，发现以下链接，希望对你有帮助。
https://blog.csdn.net/wandererXX/article/details/136935596