Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/TencentARC/MindOmni
/ functions
Functions
516 in github.com/TencentARC/MindOmni
⨍
Functions
516
◇
Types & classes
89
↳
Endpoints
2
Method
__init__
(self, config)
src/image_decoder/modeling_phi3.py:1417
Method
__init__
(self, config: Phi3Config)
src/image_decoder/modeling_phi3.py:1533
Method
__init__
(self, num_tokens_for_img: int, offload_kv_cache: bool = False)
src/image_decoder/scheduler.py:11
Method
__init__
(self, num_steps: int = 50, time_shifting_factor: int = 1)
src/image_decoder/scheduler.py:111
Method
__init__
(self, max_image_size: int = 1024)
src/image_decoder/processor.py:42
Method
__init__
(self, pad_token_id=2, llm_pad_token_id=151643, hidden_size=3072)
src/image_decoder/processor.py:90
Method
__init__
( self, vae: AutoencoderKL, model: OmniGen, connector: nn.Module, proc
src/image_decoder/image_pipeline.py:19
Method
__init__
(self, hidden_size, frequency_embedding_size=256)
src/image_decoder/model.py:23
Method
__init__
(self, hidden_size, patch_size, out_channels)
src/image_decoder/model.py:63
Method
__init__
( self, patch_size: int = 2, in_chans: int = 4, embed_dim: int
src/image_decoder/model.py:134
Method
__init__
Phi3RMSNorm is equivalent to T5LayerNorm
pretrain/modeling_phi3.py:116
Method
__init__
(self, dim, max_position_embeddings=2048, base=10000, device=None)
pretrain/modeling_phi3.py:137
Method
__init__
(self, dim, config, device=None)
pretrain/modeling_phi3.py:166
Method
__init__
(self, dim, config, device=None)
pretrain/modeling_phi3.py:207
Method
__init__
(self, dim, config, device=None)
pretrain/modeling_phi3.py:252
Method
__init__
(self, config: Phi3Config, layer_idx: Optional[int] = None)
pretrain/modeling_phi3.py:363
Method
__init__
(self, *args, **kwargs)
pretrain/modeling_phi3.py:493
Method
__init__
(self, config: Phi3Config, layer_idx: int)
pretrain/modeling_phi3.py:738
Method
__init__
(self, config: Phi3Config)
pretrain/modeling_phi3.py:950
Method
__init__
(self, config)
pretrain/modeling_phi3.py:1168
Method
__init__
(self, config)
pretrain/modeling_phi3.py:1417
Method
__init__
(self, config: Phi3Config)
pretrain/modeling_phi3.py:1533
Method
__init__
(self, num_tokens_for_img: int, offload_kv_cache: bool=False)
pretrain/scheduler.py:12
Method
__init__
(self, num_steps: int=50, time_shifting_factor: int=1)
pretrain/scheduler.py:131
Method
__init__
( self, vae: AutoencoderKL, model: OmniGen, processor: OmniGenProcessor,
pretrain/pipeline.py:46
Method
__init__
(self, text_tokenizer, max_image_size: int=1024)
pretrain/processor.py:27
Method
__init__
(self, pad_token_id=2, llm_pad_token_id=151643, hidden_size=3072)
pretrain/processor.py:351
Method
__init__
(self, hidden_size, frequency_embedding_size=256)
pretrain/model.py:28
Method
__init__
(self, hidden_size, patch_size, out_channels)
pretrain/model.py:68
Method
__init__
(self, input_dim, output_dim)
pretrain/model.py:137
Method
__init__
( self, patch_size: int = 2, in_chans: int = 4, embed_dim: int
pretrain/model.py:149
Method
__init__
( self, json_file: str, image_path: str, processer: OmniGenProcessor,
pretrain/train_helper/concatedataset.py:5
Method
__init__
Initialize the dataset.
pretrain/train_helper/webdataset_laion.py:19
Method
__init__
Initialize the dataset.
pretrain/train_helper/webdataset_.py:19
Method
__init__
( self, json_file: str, image_path: str, processer: OmniGenProcessor,
pretrain/train_helper/subdata.py:23
Method
__init__
(self, pad_token_id: int, llm_pad_token_id: int, hidden_size: int, keep_raw_resolution: bool)
pretrain/train_helper/subdata.py:133
Method
__init__
( self, json_file: str, image_path: str, processer: OmniGenProcessor,
pretrain/train_helper/data.py:21
Method
__init__
(self, pad_token_id: int, llm_pad_token_id: int, hidden_size: int, keep_raw_resolution: bool)
pretrain/train_helper/data.py:193
Method
__iter__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:105
Method
__iter__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:115
Method
__len__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/sft.py:133
Method
__len__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_rec.py:148
Method
__len__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:113
Method
__len__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:126
Method
__len__
(self)
pretrain/train_helper/concatedataset.py:105
Method
__len__
(self)
pretrain/train_helper/webdataset_laion.py:88
Method
__len__
(self)
pretrain/train_helper/webdataset_.py:86
Method
__len__
(self)
pretrain/train_helper/subdata.py:124
Method
__len__
(self)
pretrain/train_helper/data.py:187
Method
__str__
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:493
Method
_basic_init
(module)
src/image_decoder/model.py:221
Method
_basic_init
(module)
pretrain/model.py:245
Method
_get_eval_sampler
Returns a sampler for evaluation.
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:1068
Method
_get_train_sampler
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:462
Method
_get_train_sampler
Returns a sampler that ensures proper data sampling for GRPO training.
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:1052
Method
_init_weights
(self, module)
src/image_decoder/modeling_phi3.py:851
Method
_init_weights
(self, module)
pretrain/modeling_phi3.py:851
Method
_prepare_inputs
( self, inputs: dict[str, Union[torch.Tensor, Any]] )
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:502
Method
_set_signature_columns_if_needed
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:453
Method
_set_signature_columns_if_needed
(self)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:544
Method
_summarize
( ap=1, iouThr=None, areaRng='all', maxDets=100 )
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:425
Method
_summarizeDets
()
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:456
Method
_summarizeKps
()
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:471
Method
_toMask
(anns, coco)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:87
Function
accuracy_reward
Reward function that checks if the completion is correct using symbolic verification, exact string matching, or fuzzy matching.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_jsonl.py:826
Function
accuracy_reward
Reward function that checks if the completion is correct using either symbolic verification or exact string matching.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo.py:62
Function
accuracy_reward
Reward function that checks if the completion is correct using symbolic verification, exact string matching, or fuzzy matching.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_ust.py:838
Method
adjust_attention_for_input_images
(self, attention_mask, image_sizes)
src/image_decoder/processor.py:155
Method
annToMask
Convert annotation which can be polygons, uncompressed RLE, or RLE to binary mask. :return: binary mask (numpy 2D array)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/coco.py:391
Function
center_crop_arr
Center cropping implementation from ADM. https://github.com/openai/guided-diffusion/blob/8fb3ad9197f16bbc40620447b2742e13458d2831/guided_diff
pretrain/utils.py:42
Function
check_hub_revision_exists
Checks if a given Hub revision exists.
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/hub.py:70
Function
collate_fn
(examples)
rl-postrain/src/open-r1-multimodal/src/open_r1/sft.py:165
Method
computeIoU
(self, imgId, catId)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:161
Method
computeOks
(self, imgId, catId)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:190
Method
compute_loss
( self, model, inputs, return_outputs=False, num_items_in_batch=None )
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:746
Method
compute_loss
(self, model, inputs, return_outputs=False, num_items_in_batch=None)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:904
Function
construct_debug_prompt_list
()
pretrain/train_helper/validate.py:207
Function
cosine_rewards
(completions, solution, **kwargs)
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_jsonl.py:713
Function
cosine_rewards
(completions, solution, **kwargs)
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_ust.py:725
Method
crop_cache
(self, cache, num_tokens_for_img)
src/image_decoder/scheduler.py:141
Method
crop_cache
(self, cache, num_tokens_for_img)
pretrain/scheduler.py:162
Method
crop_kv_cache
(self, past_key_values, num_tokens_for_img)
src/image_decoder/scheduler.py:119
Method
crop_kv_cache
(self, past_key_values, num_tokens_for_img)
pretrain/scheduler.py:140
Method
data_collator
(features)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/vllm_grpo_trainer.py:277
Method
data_collator
(features)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:377
Method
download
Download COCO images from mscoco.org server. :param tarDir (str): COCO results directory name imgIds (list): images to
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/coco.py:323
Method
eval_per_data
(prompt, gt_prompt)
rl-postrain/src/open-r1-multimodal/src/open_r1/vlm_modules/qwen_module.py:209
Method
evaluateImg
perform evaluation for single category and image :return: dict (single image results)
rl-postrain/src/open-r1-multimodal/src/open_r1/utils/pycocotools/cocoeval.py:233
Method
extra_repr
(self)
src/image_decoder/modeling_phi3.py:131
Method
extra_repr
(self)
pretrain/modeling_phi3.py:131
Method
find_all_linear_names
(model, multimodal_keywords)
rl-postrain/src/open-r1-multimodal/src/open_r1/trainer/grpo_trainer.py:275
Function
focal_patch_weight
(pred, target, tau=1.0, gamma=2.0)
pretrain/train_helper/loss.py:150
Function
format_reward
Reward function that checks if the completion has a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_jsonl.py:882
Function
format_reward
Reward function that checks if the completion has a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo.py:105
Function
format_reward
Reward function that checks if the completion has a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/grpo_ust.py:894
Method
format_reward_rec
Check if the InternVL model output matches a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/vlm_modules/internvl_module.py:141
Method
format_reward_rec
Check if the Qwen model output matches a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/vlm_modules/qwen_module.py:103
Method
format_reward_ust
Check if the Qwen model output matches a specific format.
rl-postrain/src/open-r1-multimodal/src/open_r1/vlm_modules/qwen_module.py:123
Function
forward
( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, pos
pretrain/modeling_qwenvl.py:8
Method
forward
( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = N
src/mllm.py:11
← previous
next →
301–400 of 516, ranked by callers