Functions — github.com/NVlabs/LongLive

CodeHub MCPcopy Create free account

hub / github.com/NVlabs/LongLive / functions

Functions963 in github.com/NVlabs/LongLive

⨍ Functions 963◇ Types & classes 175↳ Endpoints 2

Functionconvert_type

fouroversix/src/fouroversix/csrc/include/utils.h:266

Methodconvert_velocity_to_x0

Convert the diffusion network's velocity prediction to x0 predidction. velocity: the predicted noise with shape [B, C, H, W]

utils/scheduler.py:77

(self, fsdp_module)

utils/distributed.py:142

Functioncp_async_wait

fouroversix/src/fouroversix/csrc/include/utils.h:284

Methodcreate_custom_forward

wan_5b/modules/causal_model.py:1591

Methodcreate_custom_forward

wan_5b/modules/model.py:520

Functioncreate_test_case

( backend_a: str = "cuda", backend_b: str = "transformer_engine", scale_rule: str = "mse", )

fouroversix/scripts/create_test_case.py:12

Methodcross_attn_ffn

(x, context, context_lens, e, crossattn_cache=None)

wan_5b/modules/causal_model.py:905

Methodcross_attn_ffn

(x, context, context_lens, e)

wan_5b/modules/model.py:272

Functioncustom_forward

(*inputs, **kwargs)

wan_5b/distributed/sequence_parallel.py:356

Methodcustom_forward

(*inputs, **kwargs)

wan_5b/modules/causal_model.py:1592

Methodcustom_forward

(*inputs, **kwargs)

wan_5b/modules/model.py:521

Methoddatabase_path

Path to the SQLite database where experiment results are stored.

fouroversix/scripts/ptq/coordinators/modal.py:25

wan_5b/modules/vae2_2.py:1037

wan_5b/modules/vae2_1.py:652

Methoddecode_to_pixel

(self, latent: torch.Tensor, use_cache: bool = False)

utils/lightvae_5b_wrapper.py:406

Methoddecode_to_pixel_chunk

Decode latent frames to pixel space. Args: latent: Latent tensor with shape [batch_size, num_frames, num_channel

utils/wan_5b_wrapper.py:211

Get device of the values in this tensor.

fouroversix/src/fouroversix/quantize/quantized_tensor.py:246

Functiondisable_sp_profiling

wan_5b/distributed/sp_ulysses_inference.py:77

wan_5b/distributed/sp_training.py:255

Functionenable_sp_profiling

wan_5b/distributed/sp_ulysses_inference.py:71

(self, videos: torch.Tensor)

inference.py:537

wan_5b/modules/vae2_2.py:1023

videos: A list of videos each with shape [C, T, H, W].

wan_5b/modules/vae2_1.py:642

Functioneval_collate_fn

Collate for text-only datasets (no frames).

utils/dataset.py:1063

Methodevaluate_on_modal

Evaluate a quantized model on Modal.

fouroversix/scripts/ptq/evaluators/evaluator.py:198

(self, prompt, system_prompt, seed=-1, *args, **kwargs)

wan_5b/utils/prompt_extend.py:158

(self, prompt, system_prompt, seed=-1, *args, **kwargs)

wan_5b/utils/prompt_extend.py:337

Methodextend_with_img

(self, prompt, system_prompt, image: U

wan_5b/utils/prompt_extend.py:194

Methodextend_with_img

(self, prompt, system_prompt, image: U

wan_5b/utils/prompt_extend.py:368

Methodextra_repr

utils/quant.py:339

Functionfake_diffusers_current_device

(model: torch.nn.Module, target_device: torch.device)

utils/memory.py:61

The filename for the kernel.

fouroversix/scripts/generate_kernels.py:75

r""" Args: x(Tensor): Shape [B, L1, C] context(Tensor): Shape [B * num_chunks, L2, C] context_lens(Tensor)

wan_5b/modules/causal_model.py:199

r""" Args: x(Tensor): Shape [B, L, num_heads, C / num_heads] seq_lens(Tensor): Shape [B] grid_sizes(Tensor

wan_5b/modules/causal_model.py:327

r""" Args: x(Tensor): Shape [B, L, C] e(Tensor): Shape [B, F, 6, C] seq_lens(Tensor): Shape [B], length of

wan_5b/modules/causal_model.py:829

r""" Args: x(Tensor): Shape [B, L1, C] e(Tensor): Shape [B, F, 1, C]

wan_5b/modules/causal_model.py:950

( self, *args, **kwargs )

wan_5b/modules/causal_model.py:1825

wan_5b/modules/t5.py:48

wan_5b/modules/t5.py:61

x: [B, L1, C]. context: [B, L2, C] or None. mask: [B, L2] or [B, L1, L2] or None.

wan_5b/modules/t5.py:86

wan_5b/modules/t5.py:136

(self, x, mask=None, pos_bias=None)

wan_5b/modules/t5.py:170

(self, x, mask=None, encoder_states=None, enco

wan_5b/modules/t5.py:206

wan_5b/modules/t5.py:233

(self, ids, mask=None)

wan_5b/modules/t5.py:301

(self, ids, mask=None, encoder_states=None, encoder_mask=None)

wan_5b/modules/t5.py:349

(self, encoder_ids, encoder_mask, decoder_ids, decoder_mask)

wan_5b/modules/t5.py:406

(self, x, seq_lens, grid_sizes, freqs, kv_cache=None, current_start=0, cache_start=None, t_sca

wan_5b/modules/causal_model_sp_ulysses.py:113

(self, x, e, seq_lens, grid_sizes, freqs, context, context_lens, kv_cache=None, crossattn_cach

wan_5b/modules/causal_model_sp_ulysses.py:342

(self, x, e, local_frame_offset=0, actual_local_frames=None, frame_seqlen_global=None)

wan_5b/modules/causal_model_sp_ulysses.py:409

(self, x, t, context, seq_len, clip_fea=None, y=None, kv_cache=None, crossattn_cache=None, cur

wan_5b/modules/causal_model_sp_ulysses.py:497

(self, x, cache_x=None)

wan_5b/modules/vae2_2.py:34

wan_5b/modules/vae2_2.py:57

Fix bfloat16 support for nearest neighbor interpolation.

wan_5b/modules/vae2_2.py:64

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_2.py:111

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_2.py:213

wan_5b/modules/vae2_2.py:254

(self, x: torch.Tensor)

wan_5b/modules/vae2_2.py:334

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_2.py:446

(self, x, feat_cache=None, feat_idx=[0], first_chunk=False)

wan_5b/modules/vae2_2.py:488

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_2.py:558

(self, x, feat_cache=None, feat_idx=[0], first_chunk=False)

wan_5b/modules/vae2_2.py:671

(self, x, scale=[0, 1])

wan_5b/modules/vae2_2.py:777

(self, x, cache_x=None)

wan_5b/modules/vae2_1.py:28

wan_5b/modules/vae2_1.py:51

Fix bfloat16 support for nearest neighbor interpolation.

wan_5b/modules/vae2_1.py:59

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_1.py:101

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_1.py:197

wan_5b/modules/vae2_1.py:235

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_1.py:313

(self, x, feat_cache=None, feat_idx=[0])

wan_5b/modules/vae2_1.py:418

r""" Args: x(Tensor): Shape [B, L, C]

wan_5b/modules/model.py:90

r""" Args: x(Tensor): Shape [B, L, C]

wan_5b/modules/model.py:106

r""" Args: x(Tensor): Shape [B, L, num_heads, C / num_heads] seq_lens(Tensor): Shape [B] grid_sizes(Tensor

wan_5b/modules/model.py:139

r""" Args: x(Tensor): Shape [B, L1, C] context(Tensor): Shape [B, L2, C] context_lens(Tensor): Shape [B]

wan_5b/modules/model.py:173

r""" Args: x(Tensor): Shape [B, L, C] e(Tensor): Shape [B, L1, 6, C] seq_lens(Tensor): Shape [B], length o

wan_5b/modules/model.py:245

( self, *args, **kwargs )

wan_5b/modules/model.py:436

(ctx, input_tensor, scatter_dim, gather_dim, group)

wan_5b/distributed/sp_training.py:137

(self, input: torch.Tensor)

utils/quant.py:335

(self, text_prompts: List[str])

utils/wan_5b_wrapper.py:43

( self, noisy_image_or_video: torch.Tensor, conditional_dict: dict, timestep: torch.Te

utils/wan_5b_wrapper.py:453

Forward pass for the FP4 MLP layer.

fouroversix/src/fouroversix/model/modules/gpt_oss.py:61

Forward pass for the FP4 experts layer.

fouroversix/src/fouroversix/model/modules/gpt_oss.py:271

Perform an FP4 matrix multiplication. The input is provided in high precision and quantized to FP4 prior to the matrix multiplication

fouroversix/src/fouroversix/model/modules/linear.py:18

Forward pass for the FP4 linear layer.

fouroversix/src/fouroversix/model/modules/linear.py:333

Perform an FP4 matrix multiplication. The input is provided in high precision and quantized to FP4 prior to the matrix multiplication

fouroversix/src/fouroversix/model/modules/linear_ori.py:18

Forward pass for the FP4 linear layer.

fouroversix/src/fouroversix/model/modules/linear_ori.py:250

Forward pass for the FP4 experts layer.

fouroversix/src/fouroversix/model/modules/qwen.py:221

Forward pass with SmoothQuant-style scaling.

fouroversix/scripts/ptq/evaluators/smoothquant.py:55

Forward pass that can optionally be run in high precision. This is used to calculate the high-precision output to compare to during t

fouroversix/scripts/ptq/evaluators/awq.py:45

( self, noisy_image_or_video: torch.Tensor, conditional_dict: dict, timestep:

pipeline/causal_diffusion_inference_sp.py:133

Functionfp4_conversion

fouroversix/src/fouroversix/csrc/include/utils.h:620

Functionfp4_dequantize_kernel

Dequantizes FP4 packed data using per-block scaling factors. Args: packed_ptr (tl.pointer): Pointer to packed uint8 tensor (M x N//2)

utils/nvfp4_kernel.py:41

Methodfp4_matmul

Perform a matrix multiplication (`a @ b.T`) between two quantized tensors.

fouroversix/src/fouroversix/matmul/pytorch.py:22

Methodfp4_matmul

Perform a matrix multiplication (`a @ b.T`) between two quantized tensors using the CUTLASS backend.

fouroversix/src/fouroversix/matmul/cutlass/backend.py:52

Functionfp4_quantization_kernel

( x_desc, x_amax_ptr, x_e2m1_desc, x_sf_desc, rbits_ptr, # Meta-parameters BLOCK_S

fouroversix/src/fouroversix/quantize/triton/kernel.py:485

Functionfree_model

wan_5b/distributed/fsdp.py:37

Functionfsdp_state_dict

utils/distributed.py:11

Functiongather_lora_state_dict

utils/lora_utils.py:77

← previous next →701–800 of 963, ranked by callers