MCPcopy Create free account

hub / github.com/LLaVA-VL/LLaVA-NeXT / functions

Functions1,515 in github.com/LLaVA-VL/LLaVA-NeXT

⨍ Functions 1,515◇ Types & classes 291↳ Endpoints 23
↓ 4 callersMethodget_placement_groups
(self, strategy: str = "STRICT_PACK", name: Optional[str] = None)
llava-critic-r1/EasyR1/verl/single_controller/ray/base.py:91
↓ 4 callersFunctionget_pretrained_cfg
(model: str, tag: str)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/pretrained.py:210
↓ 4 callersMethodget_resource_pool
Get the resource pool of the worker.
llava-critic-r1/EasyR1/verl/trainer/ray_trainer.py:98
↓ 4 callersMethodinit_model
(self)
llava-critic-r1/EasyR1/verl/workers/fsdp_workers.py:344
↓ 4 callersFunctionis_npu_available
Checks if `torch_npu` is installed and potentially if a NPU is in the environment
trl/import_utils.py:100
↓ 4 callersFunctionis_rich_available
()
trl/import_utils.py:76
↓ 4 callersMethodmerge
(self, other)
llava-critic-r1/EasyR1/verl/utils/seqlen_balancing.py:33
↓ 4 callersFunctionpad_to_length
(tensor: torch.Tensor, length: int, pad_value: Union[int, float], dim: int = -1)
trl/trainer/utils.py:531
↓ 4 callersFunctionreduce_metrics
(metrics: Dict[str, List[Any]])
llava-critic-r1/EasyR1/verl/trainer/metrics.py:23
↓ 4 callersFunctionshould_only_save_mm_adapter
(args)
llava/train/llava_trainer.py:51
↓ 4 callersMethodto_dict
(self)
trl/trainer/ppo_config.py:171
↓ 3 callersMethod__init__
(self)
llava-critic-r1/EasyR1/verl/single_controller/ray/base.py:475
↓ 3 callersMethod_get_current_device
r""" Get the current device. For GPU, we return the local process index using the `accelerate.PartialState` object to handle corner ca
trl/models/modeling_base.py:366
↓ 3 callersMethod_get_item
(self, i)
llava/train/train.py:1137
↓ 3 callersMethod_init_config
( self, config: Union[ActorConfig, CriticConfig, RefConfig], role: Literal["actor", "critic", "ref"]
llava-critic-r1/EasyR1/verl/workers/fsdp_workers.py:113
↓ 3 callersFunction_left_broadcast
As opposed to the default direction of broadcasting (right to left), this function broadcasts from left to right Args: in
trl/models/modeling_sd_base.py:156
↓ 3 callersMethod_maybe_log_save_evaluate
(self)
trl/trainer/iterative_sft_trainer.py:313
↓ 3 callersFunction_repeat_interleave
(value: Union[torch.Tensor, np.ndarray], repeats: int)
llava-critic-r1/EasyR1/verl/workers/rollout/vllm_rollout_spmd.py:34
↓ 3 callersMethod_save
(self, output_dir: Optional[str] = None, state_dict=None)
llava/train/llava_trainer.py:686
↓ 3 callersMethod_split_kwargs
Separate the kwargs from the arguments that we support inside `supported_args` and the ones that we don't.
trl/models/modeling_base.py:384
↓ 3 callersMethod_validate
(self)
llava-critic-r1/EasyR1/verl/trainer/ray_trainer.py:274
↓ 3 callersMethodappend_segment
Append a new segment to the history. args: text (`str`): The text of the new segment. tokens (`torch.LongTen
trl/environment/base_environment.py:88
↓ 3 callersMethodattention
(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/transformer.py:442
↓ 3 callersMethodautocast
Returns the autocast context manager
trl/models/modeling_sd_base.py:119
↓ 3 callersMethodbatched_forward_pass
Calculate model outputs in multiple batches. Args: queries (`torch.LongTensor`): List of tensors contain
trl/trainer/ppo_trainer.py:899
↓ 3 callersFunctionconvert_dict_to_str
(data: Dict[str, Any])
llava-critic-r1/EasyR1/verl/utils/py_functional.py:104
↓ 3 callersMethoddevice
(self)
llava/model/multimodal_encoder/eva_clip/eva_vit.py:855
↓ 3 callersFunctiondisable_torch_init
Disable the redundant torch default initialization to accelerate model creation.
llava/utils.py:163
↓ 3 callersFunctiondownload_pretrained
( cfg: Dict, force_hf_hub: bool = False, cache_dir: Union[str, None] = None, )
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/pretrained.py:286
↓ 3 callersMethodfeature_select
(self, image_forward_outs)
llava/model/multimodal_encoder/mlcd_encoder.py:51
↓ 3 callersMethodfeature_select
(self, image_forward_outs)
llava/model/multimodal_encoder/clip_encoder.py:46
↓ 3 callersMethodfinish
(self)
llava-critic-r1/EasyR1/verl/utils/logger/logger.py:51
↓ 3 callersMethodforward
(self, x)
llava/model/language_model/modeling_llama.py:212
↓ 3 callersFunctionfreeze_batch_norm_2d
Converts all `BatchNorm2d` and `SyncBatchNorm` layers of provided module into `FrozenBatchNorm2d`. If `module` is itself an instance of eithe
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/utils.py:233
↓ 3 callersFunctiongather_seq_scatter_heads
A func to sync embedding input with alltoall in sequence parallel gather sequence dimension and scatter head dim: e.g. seq_dim: 1, head_d
llava-critic-r1/EasyR1/verl/utils/ulysses.py:63
↓ 3 callersMethodgenerate_sequences
(self, prompts: DataProto)
llava-critic-r1/EasyR1/verl/workers/fsdp_workers.py:534
↓ 3 callersMethodget_2dPool
(self, image_feature, stride=2)
llava/model/llava_arch.py:171
↓ 3 callersFunctionget_length_grouped_indices
Return a list of indices so that each slice of `batch_size` consecutive indices correspond to elements of similar lengths. To do this, the in
llava/train/llava_trainer.py:145
↓ 3 callersFunctionget_model_config
(model_name)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/factory.py:70
↓ 3 callersFunctionget_processor
Create a huggingface pretrained processor.
llava-critic-r1/EasyR1/verl/utils/tokenizer.py:40
↓ 3 callersFunctionget_ulysses_sequence_parallel_world_size
Get ulysses sequence parallel world size.
llava-critic-r1/EasyR1/verl/utils/ulysses.py:47
↓ 3 callersMethodget_worker_status
(self, worker_name: str)
llava/serve/controller.py:85
↓ 3 callersFunctionimage_transform
( image_size: int, is_train: bool, mean: Optional[Tuple[float, ...]] = None, std: Optional[Tup
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/transform.py:59
↓ 3 callersMethodinitialize_vision_modules
(self, model_args, fsdp=None)
llava/model/llava_arch.py:54
↓ 3 callersMethodinitialize_vision_tokenizer
(self, model_args, tokenizer)
llava/model/llava_arch.py:557
↓ 3 callersFunctionis_diffusers_available
()
trl/import_utils.py:61
↓ 3 callersFunctionis_wandb_available
()
trl/import_utils.py:80
↓ 3 callersFunctionload_data
(data_path)
llava/train/train_dpo.py:900
↓ 3 callersMethodload_model
(self, device_map=None)
llava/model/multimodal_encoder/mlcd_encoder.py:39
↓ 3 callersMethodload_model
(self, device_map=None)
llava/model/multimodal_encoder/siglip_encoder.py:563
↓ 3 callersMethodload_model
(self, device_map="auto")
llava/model/multimodal_encoder/open_clip_encoder.py:40
↓ 3 callersMethodload_model
(self, device_map=None)
llava/model/multimodal_encoder/clip_encoder.py:35
↓ 3 callersMethodload_model
(self)
llava/model/multimodal_encoder/hf_vision.py:23
↓ 3 callersMethodload_model
(self)
llava/model/multimodal_encoder/dev_eva_clip/eva_vit.py:52
↓ 3 callersMethodload_model
(self, device_map=None)
llava/model/multimodal_encoder/eva_clip/eva_clip_encoder.py:33
↓ 3 callersMethodlock
lock modules Args: unlocked_groups (int): leave last n layer groups unlocked (default: 0)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/timm_model.py:73
↓ 3 callersMethodlog
(self, samples: List[Tuple[str, str, str, float]], step: int)
llava-critic-r1/EasyR1/verl/utils/logger/gen_logger.py:34
↓ 3 callersFunctionlogprobs_from_logits
See: https://github.com/pytorch/pytorch/issues/563#issuecomment-330103591
trl/core.py:126
↓ 3 callersFunctionmasked_mean
Compute mean of tensor with a masked values.
llava-critic-r1/EasyR1/verl/utils/torch_functional.py:72
↓ 3 callersFunctionmasked_var
Compute variance of tensor with masked values.
trl/core.py:155
↓ 3 callersFunctionmaybe_zero_3
(param, ignore_status=False, name=None)
llava/train/train.py:190
↓ 3 callersFunctionmaybe_zero_3
(param, ignore_status=False, name=None)
llava/train/train_dpo.py:171
↓ 3 callersMethodnull_ref_context
Context manager for handling null reference model (that is, peft adapter manipulation).
trl/trainer/dpo_trainer.py:629
↓ 3 callersMethodpad_sequence
(self, input_ids, batch_first, padding_value)
llava/train/train.py:1251
↓ 3 callersFunctionpad_sequence_to_length
Pad a nD tensors in the last dim to max_seq_len.
llava-critic-r1/EasyR1/verl/utils/torch_functional.py:140
↓ 3 callersFunctionprocess_anyres_image
Process an image with variable resolutions. Args: image (PIL.Image.Image): The input image to be processed. processor: The i
llava/mm_utils.py:243
↓ 3 callersFunctionprocess_highres_image
(image, processor, grid_pinpoints)
llava/mm_utils.py:98
↓ 3 callersFunctionprocess_highres_image_crop_split
(image, data_args, processor=None)
llava/mm_utils.py:87
↓ 3 callersFunctionprocess_image
(image: Union[Dict[str, Any], ImageObject, str], min_pixels: int, max_pixels: int)
llava-critic-r1/EasyR1/verl/utils/dataset.py:56
↓ 3 callersMethodprocess_image
(self, image_file, overwrite_image_aspect_ratio=None)
llava/train/train.py:1063
↓ 3 callersFunctionprocess_images
(images, image_processor, model_cfg)
llava/mm_utils.py:314
↓ 3 callersFunctionrotate_half
(x)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/rope.py:25
↓ 3 callersFunctionsave_progress
(progress_file, progress_data)
playground/sgl_llava_inference_multinode.py:26
↓ 3 callersFunctionset_seed
Helper function for reproducible behavior to set the seed in `random`, `numpy`, and `torch`. Args: seed (`int`): The seed to set.
trl/core.py:235
↓ 3 callersMethodto_dict
(self)
llava-critic-r1/EasyR1/verl/trainer/config.py:122
↓ 3 callersMethodtrain
(self, *args, **kwargs)
trl/trainer/sft_trainer.py:295
↓ 3 callersFunctionulysses_pad_and_slice_inputs
Pad and slice input_ids to be divisible by sp_size Pad position_ids to be divisible by sp_size. Note both input_ids_rmpad and position_i
llava-critic-r1/EasyR1/verl/utils/ulysses.py:262
↓ 3 callersMethodunet
Returns the 2d U-Net model used for diffusion.
trl/models/modeling_sd_base.py:84
↓ 3 callersFunctionvote_last_response
(state, vote_type, model_selector, request: gr.Request)
llava/serve/gradio_multi_image.py:84
↓ 3 callersFunctionvote_last_response
(state, vote_type, model_selector, request: gr.Request)
llava/serve/gradio_web_server.py:77
↓ 3 callersMethodzo_perturb_parameters
Perturb the model parameters with a random vector z. Args: scaling_factor (float): Scaling factor for the perturbation.
llava/train/llava_trainer.py:300
↓ 2 callersMethod__init__
(self, *, dim, dim_head=64, heads=8)
llava/model/multimodal_resampler/perceiver.py:31
↓ 2 callersMethod__init__
(self, inplanes, planes, stride=1)
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/modified_resnet.py:13
↓ 2 callersMethod__init__
( self, embed_dim: int, vision_cfg: CLIPVisionCfg, text_cfg: CLIPTextCfg,
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/model.py:222
↓ 2 callersMethod__init__
(self, config, **kwargs)
trl/models/modeling_value_head.py:26
↓ 2 callersMethod_build_messages
(self, example: Dict[str, Any])
llava-critic-r1/EasyR1/verl/utils/dataset.py:147
↓ 2 callersFunction_build_text_tower
( embed_dim: int, text_cfg: CLIPTextCfg, quick_gelu: bool = False, cast_dtype: Optional[torch.
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/model.py:190
↓ 2 callersFunction_build_vision_tower
(embed_dim: int, vision_cfg: CLIPVisionCfg, quick_gelu: bool = False, cast_dtype: Optional[torch.dtype] = None
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/model.py:125
↓ 2 callersMethod_forward_micro_batch
(self, micro_batch: Dict[str, torch.Tensor])
llava-critic-r1/EasyR1/verl/workers/critic/dp_critic.py:52
↓ 2 callersMethod_forward_micro_batch
Returns: log_probs: # (bs, response_len)
llava-critic-r1/EasyR1/verl/workers/actor/dp_actor.py:65
↓ 2 callersMethod_generate_batched
( self, model: PreTrainedModelWrapper, query_tensors: List[torch.Tensor], leng
trl/trainer/ppo_trainer.py:478
↓ 2 callersMethod_get_checkpoint_from_hub
( cls, pretrained_model, pretrained_model_name_or_path, index_filename,
trl/models/modeling_base.py:317
↓ 2 callersMethod_get_item
(self, i)
llava/train/train_dpo.py:1081
↓ 2 callersMethod_get_train_sampler
(self)
llava/train/llava_trainer.py:494
↓ 2 callersFunction_get_unpad_data
(attention_mask)
llava/model/language_model/modeling_llama.py:61
↓ 2 callersFunction_pad_tensor
(x: Tensor, dim: int, padding_size: int)
llava-critic-r1/EasyR1/verl/utils/ulysses.py:105
↓ 2 callersMethod_prepare_dataset
( self, dataset, tokenizer, packing, dataset_text_field, max_s
trl/trainer/sft_trainer.py:326
↓ 2 callersMethod_remove_unused_columns
(self, dataset: "Dataset")
trl/trainer/ppo_trainer.py:386
↓ 2 callersFunction_rescan_model_configs
()
llava/model/multimodal_encoder/dev_eva_clip/eva_clip/factory.py:33
↓ 2 callersFunction_rescan_model_configs
()
llava/model/multimodal_encoder/eva_clip/factory.py:19
← previousnext →101–200 of 1,515, ranked by callers