| from transformers.models.qwen2_5_vl.modeling_qwen2_5_vl import ( |
| Qwen2_5_VisionPatchEmbed, |
| Qwen2_5_VisionRotaryEmbedding, |
| Qwen2_5_VLVisionBlock, |
| Qwen2_5_VLPatchMerger, |
| Qwen2_5_VLPreTrainedModel |
| ) |
| from transformers.models.qwen2_5_vl.configuration_qwen2_5_vl import Qwen2_5_VLVisionConfig |
| import torch |
| import torch.nn as nn |
| import torch.nn.functional as F |
| import numpy as np |
| import transformers.models.qwen2_5_vl.modeling_qwen2_5_vl |
|
|
| def replace_qwen2_5_vision(): |
| transformers.models.qwen2_5_vl.modeling_qwen2_5_vl.Qwen2_5_VisionTransformerPretrainedModel = Qwen2_5_VisionTransformerPretrainedModelWithPatchedWindow |
|
|
| class Qwen2_5_VisionTransformerPretrainedModelWithPatchedWindow(Qwen2_5_VLPreTrainedModel): |
| config: Qwen2_5_VLVisionConfig |
| _no_split_modules = ["Qwen2_5_VLVisionBlock"] |
|
|
| def __init__(self, config, *inputs, **kwargs) -> None: |
| super().__init__(config, *inputs, **kwargs) |
| self.spatial_merge_size = config.spatial_merge_size |
| self.patch_size = config.patch_size |
| self.fullatt_block_indexes = config.fullatt_block_indexes |
| self.window_size = config.window_size |
| self.spatial_merge_unit = self.spatial_merge_size * self.spatial_merge_size |
|
|
| self.patch_embed = Qwen2_5_VisionPatchEmbed( |
| patch_size=config.patch_size, |
| temporal_patch_size=config.temporal_patch_size, |
| in_channels=config.in_channels, |
| embed_dim=config.hidden_size, |
| ) |
|
|
| head_dim = config.hidden_size // config.num_heads |
| self.rotary_pos_emb = Qwen2_5_VisionRotaryEmbedding(head_dim // 2) |
|
|
| self.blocks = nn.ModuleList( |
| [Qwen2_5_VLVisionBlock(config, config._attn_implementation) for _ in range(config.depth)] |
| ) |
| self.merger = Qwen2_5_VLPatchMerger( |
| dim=config.out_hidden_size, |
| context_dim=config.hidden_size, |
| spatial_merge_size=config.spatial_merge_size, |
| ) |
| self.gradient_checkpointing = False |
|
|
| def rot_pos_emb(self, grid_thw): |
| pos_ids = [] |
| for t, h, w in grid_thw: |
| hpos_ids = torch.arange(h).unsqueeze(1).expand(-1, w) |
| hpos_ids = hpos_ids.reshape( |
| h // self.spatial_merge_size, |
| self.spatial_merge_size, |
| w // self.spatial_merge_size, |
| self.spatial_merge_size, |
| ) |
| hpos_ids = hpos_ids.permute(0, 2, 1, 3) |
| hpos_ids = hpos_ids.flatten() |
|
|
| wpos_ids = torch.arange(w).unsqueeze(0).expand(h, -1) |
| wpos_ids = wpos_ids.reshape( |
| h // self.spatial_merge_size, |
| self.spatial_merge_size, |
| w // self.spatial_merge_size, |
| self.spatial_merge_size, |
| ) |
| wpos_ids = wpos_ids.permute(0, 2, 1, 3) |
| wpos_ids = wpos_ids.flatten() |
| pos_ids.append(torch.stack([hpos_ids, wpos_ids], dim=-1).repeat(t, 1)) |
| pos_ids = torch.cat(pos_ids, dim=0) |
| max_grid_size = grid_thw[:, 1:].max() |
| rotary_pos_emb_full = self.rotary_pos_emb(max_grid_size) |
| rotary_pos_emb = rotary_pos_emb_full[pos_ids].flatten(1) |
| return rotary_pos_emb |
|
|
| def get_window_index(self, grid_thw): |
| window_index: list = [] |
| cu_window_seqlens: list = [0] |
| window_index_id = 0 |
| vit_merger_window_size = self.window_size // self.spatial_merge_size // self.patch_size |
|
|
| for grid_t, grid_h, grid_w in grid_thw: |
| llm_grid_h, llm_grid_w = ( |
| grid_h // self.spatial_merge_size, |
| grid_w // self.spatial_merge_size, |
| ) |
| index = torch.arange(grid_t * llm_grid_h * llm_grid_w).reshape(grid_t, llm_grid_h, llm_grid_w) |
|
|
| pad_h = (vit_merger_window_size - llm_grid_h % vit_merger_window_size) % vit_merger_window_size |
| pad_w = (vit_merger_window_size - llm_grid_w % vit_merger_window_size) % vit_merger_window_size |
|
|
| num_windows_h = (llm_grid_h + pad_h) // vit_merger_window_size |
| num_windows_w = (llm_grid_w + pad_w) // vit_merger_window_size |
|
|
| index_padded = F.pad(index, (0, pad_w, 0, pad_h), "constant", -100) |
| index_padded = index_padded.reshape( |
| grid_t, |
| num_windows_h, |
| vit_merger_window_size, |
| num_windows_w, |
| vit_merger_window_size, |
| ) |
| index_padded = index_padded.permute(0, 1, 3, 2, 4).reshape( |
| grid_t, |
| num_windows_h * num_windows_w, |
| vit_merger_window_size, |
| vit_merger_window_size, |
| ) |
| seqlens = (index_padded != -100).sum([2, 3]).reshape(-1) |
| index_padded = index_padded.reshape(-1) |
| index_new = index_padded[index_padded != -100] |
| window_index.append(index_new + window_index_id) |
| cu_seqlens_tmp = seqlens.cumsum(0) * self.spatial_merge_unit + cu_window_seqlens[-1] |
| cu_window_seqlens.extend(cu_seqlens_tmp.tolist()) |
| window_index_id += (grid_t * llm_grid_h * llm_grid_w).item() |
| window_index = torch.cat(window_index, dim=0) |
|
|
| return window_index, cu_window_seqlens |
|
|
|
|
| def forward(self, hidden_states: torch.Tensor, grid_thw: torch.Tensor) -> torch.Tensor: |
| hidden_states = self.patch_embed(hidden_states) |
| seq_len, dim = hidden_states.size() |
|
|
| rotary_pos_emb = self.rot_pos_emb(grid_thw) |
|
|
| window_index, cu_window_seqlens_list = self.get_window_index(grid_thw) |
|
|
| cu_window_seqlens = torch.tensor( |
| cu_window_seqlens_list, |
| device=hidden_states.device, |
| dtype=grid_thw.dtype if torch.jit.is_tracing() else torch.int32, |
| ) |
| cu_window_seqlens = torch.unique_consecutive(cu_window_seqlens) |
|
|
| group = self.spatial_merge_unit |
| G = seq_len // group |
|
|
| hidden_states = hidden_states.view(G, group, dim) |
| rotary_pos_emb = rotary_pos_emb.view(G, group, -1) |
|
|
| window_index_dev = window_index.to(hidden_states.device, non_blocking=True) |
|
|
| hidden_states = hidden_states.index_select(0, window_index_dev).reshape(seq_len, dim) |
| rotary_pos_emb = rotary_pos_emb.index_select(0, window_index_dev).reshape(seq_len, -1) |
|
|
| emb = torch.cat((rotary_pos_emb, rotary_pos_emb), dim=-1) |
| position_embeddings = (emb.cos(), emb.sin()) |
|
|
| cu_seqlens = torch.repeat_interleave(grid_thw[:, 1] * grid_thw[:, 2], grid_thw[:, 0]).cumsum( |
| dim=0, |
| dtype=grid_thw.dtype if torch.jit.is_tracing() else torch.int32, |
| ) |
| cu_seqlens = F.pad(cu_seqlens, (1, 0), value=0) |
|
|
| if cu_seqlens.device != hidden_states.device: |
| cu_seqlens = cu_seqlens.to(hidden_states.device, non_blocking=True) |
| |
| for layer_num, blk in enumerate(self.blocks): |
| if layer_num in self.fullatt_block_indexes: |
| cu_seqlens_now = cu_seqlens |
| else: |
| cu_seqlens_now = cu_window_seqlens |
|
|
| if self.gradient_checkpointing and self.training: |
| hidden_states = self._gradient_checkpointing_func( |
| blk.__call__, hidden_states, cu_seqlens_now, None, position_embeddings |
| ) |
| else: |
| hidden_states = blk(hidden_states, cu_seqlens=cu_seqlens_now, position_embeddings=position_embeddings) |
|
|
| hidden_states = self.merger(hidden_states) |
|
|
| reverse_indices = torch.empty_like(window_index_dev) |
| reverse_indices.scatter_(0, window_index_dev, torch.arange(window_index_dev.numel(), dtype=torch.long, device=window_index_dev.device)) |
| hidden_states = hidden_states.index_select(0, reverse_indices) |
|
|
| return hidden_states |
