Functions — github.com/apple/ml-streambridge

CodeHub MCPcopy Create free account

hub / github.com/apple/ml-streambridge / functions

Functions393 in github.com/apple/ml-streambridge

⨍ Functions 393◇ Types & classes 89

↓ 1 callersFunctionevaluaet_videomme

eval/metric_report.py:134

↓ 1 callersFunctionevaluate_ovo

eval/metric_report.py:25

↓ 1 callersFunctionexpand2square

(pil_img, background_color=tuple(int(x*255) for x in OPENAI_DATASET_MEAN))

eval/utils.py:100

↓ 1 callersMethodforward

(self, hidden_state)

streambridge/model/multimodal_encoder/qwen2_vit.py:513

↓ 1 callersMethodforward_features

(self, x: torch.Tensor)

streambridge/model/multimodal_encoder/oryx_vit.py:591

↓ 1 callersMethodforward_features_list

streambridge/model/multimodal_encoder/oryx_vit.py:550

↓ 1 callersMethodforward_func

(self, images, force_fix_size=False, cal_attn_pool=False)

streambridge/model/multimodal_encoder/oryx_vit.py:798

↓ 1 callersMethodforward_llm_efficient

(self, input_ids, attention_mask, position_ids, past_key_values, inputs_embeds, labels, use_cache, output_atte

streambridge/model/language_model/qwen2vl_qwen.py:94

↓ 1 callersMethodforward_llm_efficient

(self, input_ids, attention_mask, position_ids, past_key_values, inputs_embeds, labels, use_cache, output_atte

streambridge/model/language_model/llava_qwen.py:115

↓ 1 callersMethodforward_llm_efficient

(self, input_ids, attention_mask, position_ids, past_key_values, inputs_embeds, labels, use_cache, output_atte

streambridge/model/language_model/oryx_qwen.py:91

↓ 1 callersMethodgenerate

( self, inputs: Optional[torch.Tensor] = None, images: Optional[torch.Tensor] = None,

streambridge/model/language_model/qwen2vl_qwen.py:143

↓ 1 callersMethodgenerate

( self, inputs: Optional[torch.Tensor] = None, images: Optional[torch.Tensor] = None,

streambridge/model/language_model/llava_qwen.py:164

↓ 1 callersMethodgenerate

( self, inputs: Optional[torch.Tensor] = None, images: Optional[torch.Tensor] = None,

streambridge/model/language_model/oryx_qwen.py:140

↓ 1 callersFunctionget_clip_indices

uniformyly sample num_frames frames between [start, end]

↓ 1 callersMethodget_images

(self, return_pil=False, return_path=False)

streambridge/conversation.py:236

↓ 1 callersMethodget_model

streambridge/model/language_model/qwen2vl_qwen.py:43

↓ 1 callersMethodget_model

streambridge/model/language_model/llava_qwen.py:65

↓ 1 callersMethodget_model

streambridge/model/language_model/oryx_qwen.py:44

↓ 1 callersFunctionget_pyav_video_duration

streambridge/video_utils.py:135

↓ 1 callersFunctionget_pyav_video_duration

eval/video_utils.py:130

↓ 1 callersFunctionget_qwen2_vl_video_stream

(video_path, ask_time, streaming_ckpt)

↓ 1 callersFunctionget_uniform_indices

在 [start, end] 区间内均匀采样 num_frames 帧

streambridge/video_utils.py:65

↓ 1 callersFunctionget_uniform_indices

(num_frames, vlen, start=0, end=None)

eval/video_utils.py:65

↓ 1 callersMethodget_vision_tower

streambridge/model/qwen2vl_arch.py:41

↓ 1 callersMethodget_vision_tower

streambridge/model/llava_arch.py:125

↓ 1 callersMethodget_vision_tower

streambridge/model/oryx_arch.py:104

↓ 1 callersMethodinit_bert

(self, vision_width, cross_attention_freq=2, truncation_side="right")

streambridge/model/multimodal_resampler/vlm_attention.py:251

↓ 1 callersFunctioninit_seeds

(seed=42, cuda_deterministic=True)

↓ 1 callersFunctioninsert_separator

streambridge/mm_utils.py:271

↓ 1 callersMethodintialize_system_prompts

(self, system_message = 'You are a helpful assistant.')

eval/streaming_models/online_oryx.py:140

↓ 1 callersMethodintialize_system_prompts

(self, system_message = 'You are a helpful assistant.')

eval/streaming_models/online_llava_ov.py:162

↓ 1 callersMethodmaybe_autocast

eval/streaming_models/online_oryx.py:107

↓ 1 callersMethodmaybe_autocast

eval/streaming_models/online_llava_ov.py:108

↓ 1 callersMethodmaybe_autocast

eval/streaming_models/online_qwen2_vl.py:96

↓ 1 callersMethodpooling_2d

(self, image_embeds, image_grid_thw)

streambridge/model/qwen2vl_arch.py:44

↓ 1 callersMethodpooling_2d

(self, image_embeds, image_grid_thw)

eval/streaming_models/online_qwen2_vl.py:135

↓ 1 callersMethodprune_heads

streambridge/model/multimodal_resampler/qformer.py:310

↓ 1 callersFunctionpts_to_secs

Converts a present time with the given time base and start_pts offset to seconds. Returns: time_in_seconds (float): The correspondin

streambridge/video_utils.py:121

↓ 1 callersFunctionpts_to_secs

Converts a present time with the given time base and start_pts offset to seconds. Returns: time_in_seconds (float): The correspondin

eval/video_utils.py:116

↓ 1 callersFunctionread_frames_decord

( video_path, num_frames, sample='rand', fix_start=None, max_num_frames=-1, client=None, clip

eval/video_utils.py:71

↓ 1 callersMethodresponse

( self, **generate_kwargs, )

eval/streaming_models/online_oryx.py:274

↓ 1 callersMethodrot_pos_emb

(self, grid_thw)

streambridge/model/multimodal_encoder/qwen2_vit.py:1059

↓ 1 callersMethodsave_attention_map

(self, attention_map)

streambridge/model/multimodal_resampler/qformer.py:166

↓ 1 callersMethodset_grad_checkpointing

(self, enable: bool = True)

streambridge/model/multimodal_encoder/oryx_vit.py:430

↓ 1 callersMethodsimple_forward

(self, x2, size2)

streambridge/model/multimodal_projector/builder.py:62

↓ 1 callersMethodtoken_generation

(self, text_q, vis_embed, long_video=False)

streambridge/model/multimodal_resampler/vlm_attention.py:290

↓ 1 callersMethodupdate_image_labels

(self, image_labels, chosen_frame_index, current_frame_num, tokens_per_frame)

streambridge/model/activate_videollm.py:189

(self, output_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs)

streambridge/mm_utils.py:311

Method__getattr__

streambridge/utils.py:94

(self, logger, log_level=logging.INFO)

streambridge/utils.py:88

(self, keywords, tokenizer, input_ids)

streambridge/mm_utils.py:300

streambridge/model/qwen2vl_arch.py:22

streambridge/model/llava_arch.py:39

streambridge/model/oryx_arch.py:26

streambridge/model/activate_videollm.py:29

(self, model_args)

streambridge/model/multimodal_resampler/masked_drop.py:12

streambridge/model/multimodal_resampler/qformer.py:65

(self, config, is_cross_attention)

streambridge/model/multimodal_resampler/qformer.py:123

streambridge/model/multimodal_resampler/qformer.py:290

(self, config, is_cross_attention=False)

streambridge/model/multimodal_resampler/qformer.py:304

streambridge/model/multimodal_resampler/qformer.py:361

streambridge/model/multimodal_resampler/qformer.py:376

(self, config, layer_num)

streambridge/model/multimodal_resampler/qformer.py:390

streambridge/model/multimodal_resampler/qformer.py:499

streambridge/model/multimodal_resampler/qformer.py:604

streambridge/model/multimodal_resampler/qformer.py:619

streambridge/model/multimodal_resampler/qformer.py:636

streambridge/model/multimodal_resampler/qformer.py:656

(self, config, add_pooling_layer=False)

streambridge/model/multimodal_resampler/qformer.py:698

streambridge/model/multimodal_resampler/qformer.py:984

streambridge/model/multimodal_resampler/qformer.py:1147

streambridge/model/multimodal_resampler/builder.py:14

(self, model_args, vision_tower)

streambridge/model/multimodal_resampler/vlm_attention.py:18

(self, model_args, vision_tower)

streambridge/model/multimodal_resampler/spatial_pool.py:11

(self, model_args, vision_tower)

streambridge/model/multimodal_resampler/perceiver.py:18

streambridge/model/language_model/qwen2vl_qwen.py:33

streambridge/model/language_model/llava_qwen.py:54

streambridge/model/language_model/oryx_qwen.py:33

streambridge/model/multimodal_projector/builder.py:12

(self, channels)

streambridge/model/multimodal_projector/builder.py:24

( self, dim: int, num_heads: int = 8, qkv_bias: bool = False, qk_norm:

streambridge/model/multimodal_encoder/oryx_vit.py:121

( self, dim: int, init_values: float = 1e-5, inplace: bool = False, )

streambridge/model/multimodal_encoder/oryx_vit.py:190

Args: img_size: Input image size. patch_size: Patch size. in_chans: Number of image input channels.

streambridge/model/multimodal_encoder/oryx_vit.py:263

(self, vision_tower, path, args, delay_load=False)

streambridge/model/multimodal_encoder/oryx_vit.py:753

( self, dim=None, max_position_embeddings=2048, base=10000, device=Non

streambridge/model/multimodal_encoder/qwen2_vit.py:113

(self, dim: int, theta: float = 10000.0)

streambridge/model/multimodal_encoder/qwen2_vit.py:275

( self, patch_size: int = 14, temporal_patch_size: int = 2, in_channels: int =

streambridge/model/multimodal_encoder/qwen2_vit.py:287

(self, dim: int, context_dim: int, spatial_merge_size: int = 2)

streambridge/model/multimodal_encoder/qwen2_vit.py:318

(self, dim: int, hidden_dim: int, hidden_act: str)

streambridge/model/multimodal_encoder/qwen2_vit.py:337

(self, dim: int, num_heads: int = 16)

streambridge/model/multimodal_encoder/qwen2_vit.py:348

(self, dim: int, num_heads: int = 16)

streambridge/model/multimodal_encoder/qwen2_vit.py:383

(self, dim: int, num_heads: int = 16)

streambridge/model/multimodal_encoder/qwen2_vit.py:406

(self, config, attn_implementation: str = "sdpa")

streambridge/model/multimodal_encoder/qwen2_vit.py:462

Qwen2RMSNorm is equivalent to T5LayerNorm

streambridge/model/multimodal_encoder/qwen2_vit.py:483

(self, config: Qwen2VLConfig, layer_idx: Optional[int] = None)

streambridge/model/multimodal_encoder/qwen2_vit.py:536

(self, *args, **kwargs)

streambridge/model/multimodal_encoder/qwen2_vit.py:663

(self, config: Qwen2VLConfig, layer_idx: int)

streambridge/model/multimodal_encoder/qwen2_vit.py:895

streambridge/model/multimodal_encoder/qwen2_vit.py:1025

(self, image_mean=(0.5, 0.5, 0.5), image_std=(0.5, 0.5, 0.5), size=(384, 384), crop_size: Dict[str, int] = Non

streambridge/model/multimodal_encoder/siglip_encoder.py:36

( self, hidden_size=1152, image_mean=(0.5, 0.5, 0.5), intermediate_size=4304,

streambridge/model/multimodal_encoder/siglip_encoder.py:74

← previous next →101–200 of 393, ranked by callers