gpu setting

[samqin123]

I've revised the setting.py as below, but when running marker or marker_single, still it works under CPU mode.
----reivsion ---
line10:
class Settings(BaseSettings):
# General
TORCH_DEVICE: Optional[str] = "cuda"
.....

---end-----

running status log:

(surya) E:\marker>marker E:\pdf E:\pdf
Loaded detection model vikp/surya_det3 on device cpu with dtype torch.float32
Loaded detection model vikp/surya_layout3 on device cpu with dtype torch.float32
Loaded reading order model vikp/surya_order on device cpu with dtype torch.float32
Loaded recognition model vikp/surya_rec2 on device cpu with dtype torch.float32
Config of the encoder: <class 'texify.model.model.VariableDonutSwinModel'> is overwritten by shared encoder config: VariableDonutSwinConfig {
"attention_probs_dropout_prob": 0.0,
"depths": [
2,
2,
14,
2
],
"drop_path_rate": 0.1,
"embed_dim": 128,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.0,
"hidden_size": 1024,
"image_size": [
420,
420
],
"initializer_range": 0.02,
"layer_norm_eps": 1e-05,
"mlp_ratio": 4.0,
"model_type": "donut-swin",
"num_channels": 3,
"num_heads": [
4,
8,
16,
32
],
"num_layers": 4,
"patch_size": 4,
"path_norm": true,
"qkv_bias": true,
"transformers_version": "4.46.0",
"use_2d_embeddings": false,
"use_absolute_embeddings": false,
"window_size": 5
}

Config of the decoder: <class 'transformers.models.mbart.modeling_mbart.MBartForCausalLM'> is overwritten by shared decoder config: MBartConfig {
"activation_dropout": 0.0,
"activation_function": "gelu",
"add_cross_attention": true,
"add_final_layer_norm": true,
"attention_dropout": 0.0,
"bos_token_id": 0,
"classifier_dropout": 0.0,
"d_model": 1024,
"decoder_attention_heads": 16,
"decoder_ffn_dim": 4096,
"decoder_layerdrop": 0.0,
"decoder_layers": 8,
"dropout": 0.1,
"encoder_attention_heads": 16,
"encoder_ffn_dim": 4096,
"encoder_layerdrop": 0.0,
"encoder_layers": 12,
"eos_token_id": 2,
"forced_eos_token_id": 2,
"init_std": 0.02,
"is_decoder": true,
"is_encoder_decoder": false,
"max_position_embeddings": 1536,
"model_type": "mbart",
"num_hidden_layers": 12,
"pad_token_id": 1,
"scale_embedding": true,
"tie_word_embeddings": false,
"transformers_version": "4.46.0",
"use_cache": true,
"vocab_size": 50000
}

Loaded texify model to cpu with torch.float32 dtype
Loaded recognition model vikp/surya_tablerec on device cpu with dtype torch.float32
Converting 1 pdfs in chunk 1/1 with 1 processes, and storing in E:\pdf
Detecting bboxes: 100%|██████████████████████████████████████████████████████████████████| 1/1 [00:07<00:00, 7.52s/it]
Detecting bboxes: 100%|██████████████████████████████████████████████████████████████████| 1/1 [00:06<00:00, 6.92s/it]
Finding reading order: 100%|█████████████████████████████████████████████████████████████| 1/1 [00:09<00:00, 9.96s/it]
Recognizing tables: 100%|████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00, 1.12s/it]
Recognizing equations: 100%|█████████████████████████████████████████████████████████████| 1/1 [00:03<00:00, 3.43s/it]
Processing PDFs: 100%|██████████████████████████████████████████████████████████████████| 1/1 [00:30<00:00, 30.32s/pdf]

[samqin123]

pip show torch is fine
(surya) E:\marker>pip show torch
Name: torch
Version: 2.5.0
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: https://pytorch.org/
Author: PyTorch Team
Author-email: [email protected]
License: BSD-3-Clause
Location: c:\programdata\miniconda3\envs\surya\lib\site-packages
Requires: filelock, fsspec, jinja2, networkx, sympy, typing-extensions
Required-by: marker-pdf, surya-ocr, texify