MCPcopy Create free account

hub / github.com/deepseek-ai/DeepSeek-OCR / functions

Functions95 in github.com/deepseek-ai/DeepSeek-OCR

⨍ Functions 95◇ Types & classes 23
↓ 5 callersMethod__init__
( self, cfg, freeze_embed=False, freeze_pre_norm=False )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:360
↓ 5 callersMethod__init__
Args: dim (int): Number of input channels. num_heads (int): Number of attention heads in each ViT block.
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:189
↓ 4 callersMethodget_hf_processor
(self, **kwargs: object)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:55
↓ 4 callersMethodtokenize_with_images
Tokenize text with <image> tags.
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:330
↓ 3 callersMethodforward
( self, x, patch_embeds )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:411
↓ 2 callersFunctionbuild_sam_vit_b
(checkpoint=None)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:481
↓ 2 callersMethodencode
(self, text: str, bos: bool = True, eos: bool = False)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:228
↓ 2 callersFunctionfind_closest_aspect_ratio
(aspect_ratio, target_ratios, width, height, image_size)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:11
↓ 2 callersFunctionget_rel_pos
Get relative positional embeddings according to the relative positions of query and key sizes. Args: q_size (int): size of qu
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:375
↓ 1 callersMethod__init__
(self, mean: Tuple[float, float, float] = (0.5, 0.5, 0.5), std: Tuple[float,
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:91
↓ 1 callersFunction_build_sam
( encoder_embed_dim, encoder_depth, encoder_num_heads, encoder_global_attn_indexes, checkp
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:491
↓ 1 callersMethod_parse_and_validate_image_input
( self, **kwargs: object)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:333
↓ 1 callersMethod_pixel_values_to_embedding
( self, pixel_values: torch.Tensor, images_crop: torch.Tensor, images_spatial_
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:364
↓ 1 callersMethod_process_image_input
( self, image_input)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:469
↓ 1 callersFunctionadd_decomposed_rel_pos
Calculate decomposed Relative Positional Embeddings from :paper:`mvitv2`. https://github.com/facebookresearch/mvit/blob/19786631e330df9f3622e
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:410
↓ 1 callersFunctionbuild_clip_l
()
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:447
↓ 1 callersFunctionclean_formula
(text)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_eval_batch.py:53
↓ 1 callersFunctioncount_tiles
(orig_width, orig_height, min_num=MIN_CROPS, max_num=MAX_CROPS, image_size=640, use_thumbnail=False)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:28
↓ 1 callersFunctiondraw_bounding_boxes
(image, refs, jdx)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:151
↓ 1 callersFunctiondraw_bounding_boxes
(image, refs)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:73
↓ 1 callersFunctiondynamic_preprocess
(image, min_num=MIN_CROPS, max_num=MAX_CROPS, image_size=640, use_thumbnail=False)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:45
↓ 1 callersFunctionextract_coordinates_and_label
(ref_text, image_width, image_height)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:138
↓ 1 callersFunctionextract_coordinates_and_label
(ref_text, image_width, image_height)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:60
↓ 1 callersFunctionget_abs_pos
(abs_pos, tgt_size)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:63
↓ 1 callersFunctionget_abs_pos
(abs_pos, tgt_size)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:19
↓ 1 callersMethodget_image_size_with_most_features
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:108
↓ 1 callersMethodget_input_embeddings
( self, input_ids: torch.Tensor, multimodal_embeddings: Optional[MultiModalEmbeddings]
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:508
↓ 1 callersMethodget_multimodal_embeddings
( self, **kwargs: object)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:498
↓ 1 callersMethodget_num_image_tokens
(self, *, image_width: int,
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:61
↓ 1 callersFunctionload_image
(image_path)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:28
↓ 1 callersFunctionpdf_to_images_high_quality
pdf2images
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:64
↓ 1 callersFunctionpil_to_pdf_img2pdf
(pil_images, output_path)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:97
↓ 1 callersFunctionprocess_image_with_refs
(image, ref_texts, jdx)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:218
↓ 1 callersFunctionprocess_image_with_refs
(image, ref_texts)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:140
↓ 1 callersMethodprocess_one
Args: prompt (str): the formatted prompt; conversations (List[Dict]): conversations with a list of messages;
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:241
↓ 1 callersFunctionquick_gelu
(x)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:102
↓ 1 callersFunctionre_match
(text)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:123
↓ 1 callersFunctionre_match
(text)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:45
↓ 1 callersFunctionre_match
(text)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_eval_batch.py:70
↓ 1 callersFunctionstream_generate
(image=None, prompt='')
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_image.py:147
↓ 1 callersFunctionwindow_partition
Partition into non-overlapping windows with padding if needed. Args: x (tensor): input tokens with [B, H, W, C]. window_size
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:326
↓ 1 callersFunctionwindow_unpartition
Window unpartition into original sequences and removing padding. Args: windows (tensor): input tokens with [B * num_windows, window_s
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:350
Method__call__
(self, input_ids: List[int], scores: torch.FloatTensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/ngram_norepeat.py:18
Method__call__
(self, pil_img: Image.Image)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:106
Method__call__
Args: prompt (str): the formatted prompt; images (List[ImageType]): the list of images; inference_mode (
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:298
Method__init__
(self, *, vllm_config: VllmConfig, prefix: str = "")
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:267
Method__init__
(self, hidden_size=1024, image_size=224, patch_size=14, num_channels=3)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:108
Method__init__
( self, cfg, dim: int, hidden_dim: int, )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:160
Method__init__
(self, cfg)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:228
Method__init__
(self, cfg, layer_id: int, multiple_of=256)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:287
Method__init__
(self, cfg)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:313
Method__init__
( self, embedding_dim: int, mlp_dim: int, act: Type[nn.Module] = nn.GELU,
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:44
Method__init__
(self, num_channels: int, eps: float = 1e-6)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:62
Method__init__
Args: img_size (int): Input image size. patch_size (int): Patch size. in_chans (int): Number of input ima
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:78
Method__init__
Args: dim (int): Number of input channels. num_heads (int): Number of attention heads. qkv_bias (bool):
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:255
Method__init__
Args: kernel_size (Tuple): kernel size of the projection layer. stride (Tuple): stride of the projection layer.
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:452
Method__init__
(self, cfg)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/build_linear.py:9
Method__init__
(self, ngram_size: int, window_size: int = 100, whitelist_token_ids: set = None)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/ngram_norepeat.py:9
Method__init__
( self, tokenizer: LlamaTokenizerFast = TOKENIZER, candidate_resolutions: Tuple[Tuple[
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:115
Method__str__
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:408
Method_cached_apply_hf_processor
( self, prompt: Union[str, list[int]], mm_data_items: MultiModalDataItems, hf_
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:231
Method_call_hf_processor
( self, prompt: str, mm_data: Mapping[str, object], mm_kwargs: Mapping[str, ob
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:154
Method_get_mm_fields_config
( self, hf_inputs: BatchFeature, hf_processor_mm_kwargs: Mapping[str, object], )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:178
Method_get_prompt_updates
( self, mm_items: MultiModalDataItems, hf_processor_mm_kwargs: Mapping[str, object],
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:190
Methodbos_id
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:217
Methodcompute_logits
( self, hidden_states: torch.Tensor, sampling_metadata: SamplingMetadata, )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:555
Methoddecode
(self, t: List[int], **kwargs)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:238
Methodeos_id
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:221
Methodforward
(self, input_ids: torch.Tensor, positions: torch.Tensor, inter
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:530
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:57
Methodforward
(self, pixel_values, patch_embeds)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:131
Methodforward
(self, x)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:171
Methodforward
( self, x: torch.Tensor, )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:243
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:305
Methodforward
( self, hidden_states, )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:329
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:55
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:68
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:169
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:233
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:291
Methodforward
(self, x: torch.Tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/sam_vary_sdpa.py:474
Methodforward
(self, x)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/build_linear.py:97
Methodget_dummy_mm_data
( self, seq_len: int, mm_counts: Mapping[str, int], )
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:126
Methodget_dummy_text
(self, mm_counts: Mapping[str, int])
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:118
Methodget_flops_per_sample
(cfg)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/build_linear.py:159
Methodget_hf_config
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:52
Methodget_language_model
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:495
Methodget_replacement_deepseek_vl2
(item_idx: int)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:201
Methodget_supported_mm_limits
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:58
Methodload_weights
(self, weights: Iterable[Tuple[str, torch.Tensor]])
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepseek_ocr.py:564
Methodpad_id
(self)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/process/image_process.py:225
Functionprocess_formula
(match)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_eval_batch.py:57
Functionprocess_single_image
single image
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_pdf.py:223
Functionprocess_single_image
single image
DeepSeek-OCR-master/DeepSeek-OCR-vllm/run_dpsk_ocr_eval_batch.py:81
Methodset_input_tensor
(self, input_tensor)
DeepSeek-OCR-master/DeepSeek-OCR-vllm/deepencoder/clip_sdpa.py:403