Skip to content
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

Fix: Qwen2-VL training on video datasets #33307

Merged
merged 4 commits into from
Sep 17, 2024
Merged

Fix: Qwen2-VL training on video datasets #33307

Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2fbec02
fix video finetuning
hiyouga Sep 4, 2024
bb4c2b6
Update modeling_qwen2_vl.py
hiyouga Sep 4, 2024
96286c3
Update modeling_qwen2_vl.py
hiyouga Sep 4, 2024
627d147
fix
hiyouga Sep 5, 2024
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
16 changes: 8 additions & 8 deletions src/transformers/models/qwen2_vl/modeling_qwen2_vl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1577,16 +1577,16 @@ def forward(
inputs_embeds = self.model.embed_tokens(input_ids)
if pixel_values is not None:
pixel_values = pixel_values.type(self.visual.get_dtype())
image_embeds = self.visual(pixel_values, grid_thw=image_grid_thw).to(inputs_embeds.device)
image_mask = input_ids == self.config.image_token_id
if self.training:
inputs_embeds = inputs_embeds.clone()
inputs_embeds[image_mask] = image_embeds
image_embeds = self.visual(pixel_values, grid_thw=image_grid_thw)
image_mask = (input_ids == self.config.image_token_id).unsqueeze(-1).expand_as(inputs_embeds)
image_embeds = image_embeds.to(inputs_embeds.device, inputs_embeds.dtype)
inputs_embeds.masked_scatter(image_mask, image_embeds)
if pixel_values_videos is not None:
pixel_values_videos = pixel_values_videos.type(self.visual.get_dtype())
video_embeds = self.visual(pixel_values_videos, grid_thw=video_grid_thw).to(inputs_embeds.device)
video_mask = input_ids == self.config.video_token_id
inputs_embeds[video_mask] = video_embeds
video_embeds = self.visual(pixel_values_videos, grid_thw=video_grid_thw)
video_mask = (input_ids == self.config.video_token_id).unsqueeze(-1).expand_as(inputs_embeds)
video_embeds = video_embeds.to(inputs_embeds.device, inputs_embeds.dtype)
inputs_embeds.masked_scatter(video_mask, video_embeds)
if attention_mask is not None:
attention_mask = attention_mask.to(inputs_embeds.device)

Expand Down
Loading