MCPcopy Create free account

hub / github.com/JIA-Lab-research/LISA / functions

Functions403 in github.com/JIA-Lab-research/LISA

⨍ Functions 403◇ Types & classes 70
↓ 1 callersMethod_apply_prefix_mask
(self, attn_bias: torch.Tensor, prefix_mask: torch.Tensor)
model/llava/model/language_model/mpt/modeling_mpt.py:169
↓ 1 callersMethod_apply_sequence_id
( self, attn_bias: torch.Tensor, sequence_id: torch.LongTensor )
model/llava/model/language_model/mpt/modeling_mpt.py:192
↓ 1 callersMethod_attn_bias
( self, device, dtype, attention_mask: Optional[torch.ByteTensor] = None,
model/llava/model/language_model/mpt/modeling_mpt.py:116
↓ 1 callersFunction_convert_bloom_causal_lm_to_prefix_lm
Converts a BLOOM Causal LM to a Prefix LM. Supported HuggingFace model classes: - `BloomForCausalLM` See `convert_hf_causal_lm_to_pr
model/llava/model/language_model/mpt/hf_prefixlm_converter.py:183
↓ 1 callersFunction_convert_gpt_causal_lm_to_prefix_lm
Converts a GPT-style Causal LM to a Prefix LM. Supported HuggingFace model classes: - `GPT2LMHeadModel` - `GPTNeoForCausalLM`
model/llava/model/language_model/mpt/hf_prefixlm_converter.py:45
↓ 1 callersFunction_convert_opt_causal_lm_to_prefix_lm
Converts an OPT Causal LM to a Prefix LM. Supported HuggingFace model classes: - `OPTForCausalLM` See `convert_hf_causal_lm_to_prefi
model/llava/model/language_model/mpt/hf_prefixlm_converter.py:531
↓ 1 callersMethod_embed_boxes
Embeds box prompts.
model/segment_anything/modeling/prompt_encoder.py:100
↓ 1 callersMethod_embed_masks
Embeds mask inputs.
model/segment_anything/modeling/prompt_encoder.py:111
↓ 1 callersMethod_embed_masks
( self, input_mask: torch.Tensor, has_mask_input: torch.Tensor )
model/segment_anything/utils/onnx.py:74
↓ 1 callersMethod_embed_points
Embeds point prompts.
model/segment_anything/modeling/prompt_encoder.py:78
↓ 1 callersMethod_embed_points
( self, point_coords: torch.Tensor, point_labels: torch.Tensor )
model/segment_anything/utils/onnx.py:51
↓ 1 callersMethod_generate_masks
(self, image: np.ndarray)
model/segment_anything/automatic_mask_generator.py:189
↓ 1 callersMethod_get_batch_fmtstr
(self, num_batches)
utils/utils.py:147
↓ 1 callersMethod_get_batch_size
Gets the batch size of the output given the batch size of the input prompts.
model/segment_anything/modeling/prompt_encoder.py:116
↓ 1 callersMethod_get_device
(self)
model/segment_anything/modeling/prompt_encoder.py:137
↓ 1 callersFunction_mask_targets
(target, tokenized_lens, speakers)
model/llava/train/train.py:281
↓ 1 callersFunction_normal_init_
(std, mean=0.0)
model/llava/model/language_model/mpt/param_init_fns.py:164
↓ 1 callersMethod_process_batch
( self, points: np.ndarray, im_size: Tuple[int, ...], crop_box: List[int],
model/segment_anything/automatic_mask_generator.py:260
↓ 1 callersMethod_process_crop
( self, image: np.ndarray, crop_box: List[int], crop_layer_idx: int, o
model/segment_anything/automatic_mask_generator.py:217
↓ 1 callersMethod_recombine_heads
(self, x: Tensor)
model/segment_anything/modeling/transformer.py:215
↓ 1 callersMethod_save
(self, output_dir: Optional[str] = None, state_dict=None)
model/llava/train/llava_trainer.py:63
↓ 1 callersFunctionadapt_tokenizer_for_denoising
Adds sentinel tokens and padding token (if missing). Expands the tokenizer vocabulary to include sentinel tokens used in mixture-of-denoiser
model/llava/model/language_model/mpt/adapt_tokenizer.py:10
↓ 1 callersFunctionadd_decomposed_rel_pos
Calculate decomposed Relative Positional Embeddings from :paper:`mvitv2`. https://github.com/facebookresearch/mvit/blob/19786631e330df9f3622e
model/segment_anything/modeling/image_encoder.py:354
↓ 1 callersMethodapply_boxes
Expects a numpy array shape Bx4. Requires the original image size in (H, W) format.
model/segment_anything/utils/transforms.py:52
↓ 1 callersMethodapply_coords_torch
Expects a torch tensor with length 2 in the last dimension. Requires the original image size in (H, W) format.
model/segment_anything/utils/transforms.py:76
↓ 1 callersFunctionapply_delta
(base_model_path, target_model_path, delta_path)
model/llava/model/apply_delta.py:13
↓ 1 callersFunctionarea_from_rle
(rle: Dict[str, Any])
model/segment_anything/utils/amg.py:152
↓ 1 callersFunctionattn_bias_shape
( attn_impl, n_heads, seq_len, alibi, prefix_lm, causal, use_sequence_id )
model/llava/model/language_model/mpt/attention.py:457
↓ 1 callersMethodbackward
(ctx, do)
model/llava/model/language_model/mpt/flash_attn_triton.py:1061
↓ 1 callersFunctionbatch_iterator
(batch_size: int, *args)
model/segment_anything/utils/amg.py:98
↓ 1 callersFunctionbuild_alibi_bias
( n_heads, seq_len, full=False, alibi_bias_max=8, device=None, dtype=None )
model/llava/model/language_model/mpt/attention.py:507
↓ 1 callersFunctionbuild_all_layer_point_grids
Generates point grids for all crop layers.
model/segment_anything/utils/amg.py:189
↓ 1 callersFunctionbuild_attn_bias
( attn_impl, attn_bias, n_heads, seq_len, causal=False, alibi=False, alibi_bias_max=8 )
model/llava/model/language_model/mpt/attention.py:474
↓ 1 callersFunctionbuild_point_grid
Generates a 2D grid of points evenly spaced in [0,1]x[0,1].
model/segment_anything/utils/amg.py:179
↓ 1 callersFunctionbuild_sam_vit_h
(checkpoint=None)
model/segment_anything/build_sam.py:15
↓ 1 callersFunctioncoco_encode_rle
(uncompressed_rle: Dict[str, Any])
model/segment_anything/utils/amg.py:294
↓ 1 callersFunctionconsolidate_ckpt
(src_path, dst_path)
model/llava/model/consolidate.py:13
↓ 1 callersMethodcreateIndex
(self)
utils/grefer.py:84
↓ 1 callersMethodcreateIndex
(self)
utils/refer.py:81
↓ 1 callersFunctioncreate_custom_forward
(module)
model/llava/model/language_model/mpt/hf_prefixlm_converter.py:367
↓ 1 callersMethoddevice
(self)
model/llava/model/multimodal_encoder/clip_encoder.py:71
↓ 1 callersFunctiondice_loss
Compute the DICE loss, similar to generalized IOU for masks Args: inputs: A float tensor of arbitrary shape. The pred
model/LISA.py:16
↓ 1 callersMethoddisplay
(self, batch)
utils/utils.py:137
↓ 1 callersFunctionfind_all_linear_names
(model)
model/llava/train/train.py:172
↓ 1 callersFunctionfind_linear_layers
(model, lora_target_modules)
train_ds.py:179
↓ 1 callersFunctionfind_linear_layers
(model, lora_target_modules)
merge_lora_weights_and_save_hf_model.py:106
↓ 1 callersFunctionfused_init_helper_
(module: nn.Module, init_fn_)
model/llava/model/language_model/mpt/param_init_fns.py:21
↓ 1 callersFunctiongen_slopes
(n_heads, alibi_bias_max=8, device=None)
model/llava/model/language_model/mpt/attention.py:497
↓ 1 callersMethodgenerate
Generates masks for the given image. Arguments: image (np.ndarray): The image to generate masks for, in HWC uint8 format.
model/segment_anything/automatic_mask_generator.py:127
↓ 1 callersFunctiongenerate_crop_boxes
Generates a list of crop boxes of different sizes. Each layer has (2**i)**2 boxes for the ith layer.
model/segment_anything/utils/amg.py:200
↓ 1 callersMethodgetAnnIds
(self, image_ids=[], ref_ids=[])
utils/grefer.py:186
↓ 1 callersMethodgetAnnIds
(self, image_ids=[], cat_ids=[], ref_ids=[])
utils/refer.py:179
↓ 1 callersMethodgetImgIds
(self, ref_ids=[])
utils/grefer.py:210
↓ 1 callersMethodgetMask
(self, ref)
utils/refer.py:308
↓ 1 callersMethodgetRefBox
(self, ref_id)
utils/grefer.py:236
↓ 1 callersMethodgetRefBox
(self, ref_id)
utils/refer.py:241
↓ 1 callersMethodgetRefIds
(self, image_ids=[], cat_ids=[], split=[])
utils/grefer.py:164
↓ 1 callersMethodget_images
(self, return_pil=False)
model/llava/conversation.py:112
↓ 1 callersFunctionget_mm_adapter_state_maybe_zero_3
(named_params, keys_to_match)
model/llava/train/train.py:160
↓ 1 callersFunctionget_mm_adapter_state_maybe_zero_3
(named_params, keys_to_match)
model/llava/train/llava_trainer.py:23
↓ 1 callersFunctionget_peft_state_maybe_zero_3
(named_params, bias)
model/llava/train/train.py:125
↓ 1 callersFunctionget_peft_state_non_lora_maybe_zero_3
(named_params, require_grad_only=True)
model/llava/train/train.py:150
↓ 1 callersFunctionget_tokenize_len
(prompts)
model/llava/train/train.py:655
↓ 1 callersMethodget_vision_tower
(self)
model/llava/model/llava_arch.py:90
↓ 1 callersFunctioninit_on_device
Device initialization context manager. A context manager under which models are initialized with all parameters on the specified device.
model/llava/model/language_model/mpt/meta_init_context.py:40
↓ 1 callersMethodinitialize_vision_tokenizer
(self, model_args, num_new_tokens)
model/llava/model/llava_arch.py:350
↓ 1 callersFunctioninsert_separator
(X, sep)
model/llava/mm_utils.py:24
↓ 1 callersFunctionintersectionAndUnionGPU
(output, target, K, ignore_index=255)
utils/utils.py:116
↓ 1 callersFunctionis_box_near_crop_edge
Filter masks at the edge of a crop, but not at the edge of the original image.
model/segment_anything/utils/amg.py:78
↓ 1 callersMethodloadAnns
(self, ann_ids=[])
utils/refer.py:223
↓ 1 callersMethodloadImgs
(self, image_ids=[])
utils/grefer.py:230
↓ 1 callersMethodloadRefs
(self, ref_ids=[])
utils/grefer.py:222
↓ 1 callersFunctionload_from_hf
(repo_id, filename, subfolder=None)
model/llava/model/builder.py:82
↓ 1 callersFunctionload_grefcoco_json
( refer_root, dataset_name, splitby, split, image_root, extra_annotation_keys=None,
utils/grefcoco.py:25
↓ 1 callersFunctionmain
(args)
train_ds.py:109
↓ 1 callersFunctionmain
(args)
chat.py:66
↓ 1 callersFunctionmain
(args)
merge_lora_weights_and_save_hf_model.py:58
↓ 1 callersFunctionmake_delta
(base_model_path, target_model_path, delta_path, hub_repo_id)
model/llava/model/make_delta.py:13
↓ 1 callersFunctionmake_supervised_data_module
Make dataset and collator for supervised fine-tuning.
model/llava/train/train.py:797
↓ 1 callersMethodmask_postprocessing
( self, masks: torch.Tensor, orig_im_size: torch.Tensor )
model/segment_anything/utils/onnx.py:85
↓ 1 callersMethodmatch_any
(a, b)
utils/grefer.py:79
↓ 1 callersFunctionmaybe_zero_3
(param, ignore_status=False, name=None)
model/llava/train/llava_trainer.py:8
↓ 1 callersMethodmodel_forward
( self, images: torch.FloatTensor, images_clip: torch.FloatTensor, input_ids:
model/LISA.py:169
↓ 1 callersFunctionparse_args
(args)
train_ds.py:24
↓ 1 callersFunctionparse_args
(args)
chat.py:19
↓ 1 callersFunctionparse_args
(args)
app.py:23
↓ 1 callersFunctionparse_args
(args)
merge_lora_weights_and_save_hf_model.py:18
↓ 1 callersFunctionpatch_tensor_constructor
(fn)
model/llava/model/language_model/mpt/meta_init_context.py:85
↓ 1 callersMethodpostprocess_small_regions
Removes small disconnected regions and holes in masks, then reruns box NMS to remove any new duplicates. Edits mask_data in
model/segment_anything/automatic_mask_generator.py:324
↓ 1 callersFunctionpreprocess
Normalize pixel values and pad to a square input.
chat.py:49
↓ 1 callersFunctionpreprocess
Normalize pixel values and pad to a square input.
app.py:53
↓ 1 callersFunctionpreprocess
Given a list of sources, each is a conversation list. This transform: 1. Add signal '### ' at the beginning each sentence, with end signal '\
model/llava/train/train.py:621
↓ 1 callersFunctionpreprocess_llama_2
( sources, tokenizer: transformers.PreTrainedTokenizer, has_image: bool = False )
model/llava/train/train.py:344
↓ 1 callersFunctionpreprocess_mpt
( sources, tokenizer: transformers.PreTrainedTokenizer, )
model/llava/train/train.py:516
↓ 1 callersFunctionpreprocess_multimodal
(source, mm_use_im_start_end)
utils/vqa_dataset.py:16
↓ 1 callersFunctionpreprocess_multimodal
(sources: Sequence[str], data_args: DataArguments)
model/llava/train/train.py:314
↓ 1 callersFunctionpreprocess_plain
( sources: Sequence[str], tokenizer: transformers.PreTrainedTokenizer, )
model/llava/train/train.py:592
↓ 1 callersFunctionpreprocess_v1
( sources, tokenizer: transformers.PreTrainedTokenizer, has_image: bool = False )
model/llava/train/train.py:430
↓ 1 callersFunctionreplace_llama_attn_with_flash_attn
()
model/llava/train/llama_flash_attn_monkey_patch.py:116
↓ 1 callersMethodresize_longest_image_size
( input_image_size: torch.Tensor, longest_side: int )
model/segment_anything/utils/onnx.py:42
← previousnext →101–200 of 403, ranked by callers