File size: 8,011 Bytes
8b306b3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
# Copyright (c) 2025 ByteDance Ltd. and/or its affiliates.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# coding: utf-8

"""
Data helpers used by inference (`inference_lance.py`, `ValidationDataset`) and the
Lance model core (`modeling/lance/lance.py`).

Exported utilities:
    - Position id helpers (image / video, interpolate / extrapolate)
    - Patchify helpers (image + video-with-merge)
    - create_sparse_mask       : flex-attention sparse mask builder
    - add_special_tokens       : register chat / vision tokens on a tokenizer
    - len2weight               : CE loss reweighting factor
"""

from einops import rearrange

import torch
from torch.nn.attention.flex_attention import or_masks, and_masks


# ------------------------------------------------------------------
# Position id helpers
# ------------------------------------------------------------------

def get_flattened_position_ids_interpolate_video(num_frames, img_h, img_w, patch_size, max_num_frames, max_num_patches_per_side):
    num_patches_h, num_patches_w = img_h // patch_size, img_w // patch_size
    # temporal
    boundaries_t = torch.arange(1 / max_num_frames, 1.0, 1 / max_num_frames)
    fractional_coords_t = torch.arange(0, 1 - 1e-6, 1 / num_frames)
    bucket_coords_t = torch.bucketize(fractional_coords_t, boundaries_t, right=True)
    # spatial
    boundaries_s = torch.arange(1 / max_num_patches_per_side, 1.0, 1 / max_num_patches_per_side)
    fractional_coords_h = torch.arange(0, 1 - 1e-6, 1 / num_patches_h)
    fractional_coords_w = torch.arange(0, 1 - 1e-6, 1 / num_patches_w)
    bucket_coords_h = torch.bucketize(fractional_coords_h, boundaries_s, right=True)
    bucket_coords_w = torch.bucketize(fractional_coords_w, boundaries_s, right=True)
    pos_ids = (
        bucket_coords_t[:, None, None] * max_num_patches_per_side * max_num_patches_per_side
        + bucket_coords_h[None, :, None] * max_num_patches_per_side
        + bucket_coords_w[None, None, :]
    ).flatten()
    return pos_ids


def get_flattened_position_ids_extrapolate_video(t, h, w, max_latent_size):
    """
    默认情况下:
        num_frames = 7 (对应 25 frames)
        max_num_patches_per_side = 64
    """
    coords_t = torch.arange(0, t)
    coords_h = torch.arange(0, h)
    coords_w = torch.arange(0, w)
    pos_ids = (
        coords_t[:, None, None] * max_latent_size * max_latent_size
        + coords_h[None, :, None] * max_latent_size
        + coords_w[None, None, :]
    ).flatten()
    return pos_ids


def get_flattened_position_ids_extrapolate(img_h, img_w, patch_size, max_num_patches_per_side):
    num_patches_h, num_patches_w = img_h // patch_size, img_w // patch_size
    coords_h = torch.arange(0, num_patches_h)
    coords_w = torch.arange(0, num_patches_w)
    pos_ids = (coords_h[:, None] * max_num_patches_per_side + coords_w).flatten()
    return pos_ids


def get_flattened_position_ids_interpolate(img_h, img_w, patch_size, max_num_patches_per_side):
    num_patches_h, num_patches_w = img_h // patch_size, img_w // patch_size
    boundaries = torch.arange(1 / max_num_patches_per_side, 1.0, 1 / max_num_patches_per_side)
    fractional_coords_h = torch.arange(0, 1 - 1e-6, 1 / num_patches_h)
    fractional_coords_w = torch.arange(0, 1 - 1e-6, 1 / num_patches_w)
    bucket_coords_h = torch.bucketize(fractional_coords_h, boundaries, right=True)
    bucket_coords_w = torch.bucketize(fractional_coords_w, boundaries, right=True)
    pos_ids = (bucket_coords_h[:, None] * max_num_patches_per_side + bucket_coords_w).flatten()
    return pos_ids


# ------------------------------------------------------------------
# Patchify helpers
# ------------------------------------------------------------------

def patchify(image, patch_size):
    p = patch_size
    c, h, w = image.shape
    assert h % p == 0 and w % p == 0
    image = image.reshape(c, h // p, p, w // p, p)
    image = torch.einsum("chpwq->hwpqc", image)
    image = image.reshape(-1, p**2 * c)
    return image


def patchify_video_with_merge(video, spatial_patch_size, temporal_patch_size, merge_size=2):
    """
    Args:
        video: Tensor of shape [C, T, H, W]
        spatial_patch_size: patch size for H/W
        temporal_patch_size: patch size for T
        merge_size: merging factor for spatial grid (固定为 2)

    Returns:
        patches: Tensor of shape [num_patches, patch_dim]
    """
    video = rearrange(video, "C T H W -> T C H W")
    T, C, H, W = video.shape
    p, tp, ms = spatial_patch_size, temporal_patch_size, merge_size

    gt, gh, gw = T // tp, H // p, W // p
    video = video.reshape(gt, tp, C, gh // ms, ms, p, gw // ms, ms, p)
    video = video.permute(0, 3, 6, 4, 7, 2, 1, 5, 8)
    patches = video.reshape(gt * gh * gw, C * tp * p * p)
    return patches


# ------------------------------------------------------------------
# Sparse attention mask (flex-attention)
# ------------------------------------------------------------------

def create_sparse_mask(document_lens, split_lens, attn_modes, device):
    def causal_mask(b, h, q_idx, kv_idx):
        return q_idx >= kv_idx

    def full_and_noise_mask(b, h, q_idx, kv_idx):
        return (full_and_noise_seq_id[q_idx] == full_and_noise_seq_id[kv_idx]) & (full_and_noise_seq_id[q_idx] >= 0)

    def remove_noise_mask(b, h, q_idx, kv_idx):
        return ~((noise_seq_id[kv_idx] >= 0) & (noise_seq_id[q_idx] != noise_seq_id[kv_idx]))

    def sample_mask(b, h, q_idx, kv_idx):
        return document_id[q_idx] == document_id[kv_idx]

    full_and_noise_tmp = []
    noise_tmp = []

    for i, (length, mode) in enumerate(zip(split_lens, attn_modes)):
        value = i if mode in ["full", "noise"] else -1
        full_and_noise_tmp.extend([value] * length)
        value_noise = i if mode == "noise" else -1
        noise_tmp.extend([value_noise] * length)

    full_and_noise_seq_id = torch.Tensor(full_and_noise_tmp).to(device)
    noise_seq_id = torch.Tensor(noise_tmp).to(device)

    document_id = torch.cat([torch.full((l,), i) for i, l in enumerate(document_lens, start=1)]).to(device)

    return and_masks(or_masks(causal_mask, full_and_noise_mask), remove_noise_mask, sample_mask)


# ------------------------------------------------------------------
# Tokenizer / loss helpers
# ------------------------------------------------------------------

def add_special_tokens(tokenizer):
    all_special_tokens = []
    for k, v in tokenizer.special_tokens_map.items():
        if isinstance(v, str):
            all_special_tokens.append(v)
        elif isinstance(v, list):
            all_special_tokens += v

    new_tokens = []
    for tok in ("<|im_start|>", "<|im_end|>", "<|vision_start|>", "<|vision_end|>"):
        if tok not in all_special_tokens:
            new_tokens.append(tok)

    num_new_tokens = tokenizer.add_tokens(new_tokens)
    new_token_ids = dict(
        bos_token_id=tokenizer.convert_tokens_to_ids("<|im_start|>"),
        eos_token_id=tokenizer.convert_tokens_to_ids("<|im_end|>"),
        start_of_image=tokenizer.convert_tokens_to_ids("<|vision_start|>"),
        end_of_image=tokenizer.convert_tokens_to_ids("<|vision_end|>"),
    )
    return tokenizer, new_token_ids, num_new_tokens


def len2weight(x, loss_reduction="square"):
    if x == 0:
        return x
    if loss_reduction == "token":
        return 1
    if loss_reduction == "sample":
        return 1 / x
    if loss_reduction == "square":
        return 1 / (x**0.5)
    raise NotImplementedError(loss_reduction)