Skip to content

add image gen #56

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
williechai wants to merge 1 commit into
base: Ming_Flash_2.0
Choose a base branch
from
Open

add image gen #56

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
67 changes: 67 additions & 0 deletions bailingmm_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,6 +41,17 @@
FPS_MIN_FRAMES = 4
FPS_MAX_FRAMES = 128

VideoInput = Union[
List["Image.Image"],
"np.ndarray",
"torch.Tensor",
List["np.ndarray"],
List["torch.Tensor"],
List[List["Image.Image"]],
List[List["np.ndarrray"]],
Copy link

Copilot AI Sep 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's a typo in 'np.ndarrray' - it should be 'np.ndarray' with only one 'r'.

Suggested change
List[List["np.ndarrray"]],
List[List["np.ndarray"]],

Copilot uses AI. Check for mistakes.
List[List["torch.Tensor"]],
]

def is_decord_available() -> bool:
import importlib.util
return importlib.util.find_spec("decord") is not None
Expand Down Expand Up @@ -504,3 +515,59 @@ def process_vision_info(
if len(audio_inputs) == 0:
audio_inputs = None
return image_inputs, video_inputs, audio_inputs

def get_closest_ratio(height: float, width: float, aspect_ratios: dict):
aspect_ratio = height / width
closest_ratio = min(aspect_ratios.keys(), key=lambda ratio: abs(float(ratio) - aspect_ratio))
return aspect_ratios[closest_ratio], float(closest_ratio)

def process_ratio(ori_h, ori_w):
ASPECT_RATIO_512 = {
"0.25": [256, 1024],
"0.26": [256, 992],
"0.27": [256, 960],
"0.28": [256, 928],
"0.32": [288, 896],
"0.33": [288, 864],
"0.35": [288, 832],
"0.4": [320, 800],
"0.42": [320, 768],
"0.48": [352, 736],
"0.5": [352, 704],
"0.52": [352, 672],
"0.57": [384, 672],
"0.6": [384, 640],
"0.68": [416, 608],
"0.72": [416, 576],
"0.78": [448, 576],
"0.82": [448, 544],
"0.88": [480, 544],
"0.94": [480, 512],
"1.0": [512, 512],
"1.07": [512, 480],
"1.13": [544, 480],
"1.21": [544, 448],
"1.29": [576, 448],
"1.38": [576, 416],
"1.46": [608, 416],
"1.67": [640, 384],
"1.75": [672, 384],
"2.0": [704, 352],
"2.09": [736, 352],
"2.4": [768, 320],
"2.5": [800, 320],
"2.89": [832, 288],
"3.0": [864, 288],
"3.11": [896, 288],
"3.62": [928, 256],
"3.75": [960, 256],
"3.88": [992, 256],
"4.0": [1024, 256],
}
closest_size, _ = get_closest_ratio(ori_h, ori_w, aspect_ratios=ASPECT_RATIO_512)
closest_size = list(map(lambda x: int(x), closest_size))
if closest_size[0] / ori_h > closest_size[1] / ori_w:
resize_size = closest_size[0], int(ori_w * closest_size[0] / ori_h)
else:
resize_size = int(ori_h * closest_size[1] / ori_w), closest_size[1]
return closest_size, resize_size
2 changes: 1 addition & 1 deletion image_processing_bailingmm2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@
resize,
to_channel_dimension_format,
)
from transformers.video_utils import VideoInput
from bailingmm_utils import VideoInput
from transformers.image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
Expand Down
13 changes: 12 additions & 1 deletion modeling_bailing_moe_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1561,6 +1561,8 @@ def forward(
output_router_logits: Optional[bool] = None,
return_dict: Optional[bool] = None,
second_per_grid_ts: Optional[torch.Tensor] = None,
image_mask=None,
audio_mask=None,
**kwargs,
) -> Union[Tuple, MoeModelOutputWithPast]:

Expand Down Expand Up @@ -1595,7 +1597,12 @@ def forward(

# retrieve input_ids and inputs_embeds
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
assert input_ids.size(1) == inputs_embeds.size(1), "{} vs {}".format(
input_ids.size,
inputs_embeds.size,
)
batch_size, seq_length = inputs_embeds.shape[:2]
#raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
Copy link

Copilot AI Sep 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Commented-out error handling should be removed or replaced with proper logic. If this validation is no longer needed, remove the comment entirely.

Suggested change
#raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")

Copilot uses AI. Check for mistakes.
elif input_ids is not None:
batch_size, seq_length = input_ids.shape[:2]
elif inputs_embeds is not None:
Expand Down Expand Up @@ -1810,6 +1817,8 @@ def forward(
return_dict: Optional[bool] = None,
second_per_grid_ts: Optional[torch.Tensor] = None,
num_logits_to_keep: Optional[int] = 0,
image_mask=None,
audio_mask=None,
**kwargs,
) -> Union[Tuple, MoeCausalLMOutputWithPast]:
r"""
Expand Down Expand Up @@ -1865,6 +1874,8 @@ def forward(
output_router_logits=output_router_logits,
return_dict=return_dict,
second_per_grid_ts=second_per_grid_ts,
image_mask=image_mask,
audio_mask=audio_mask,
**kwargs,
)

Expand Down
Loading