anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

[Refactor] Use trajectory_* fields for Qwen-Image structured RL outputs (#2513)

Browse Source 
Signed-off-by: samithuang <285365963@qq.com>
This commit is contained in:
Samit authored and GitHub committed 2026-04-09 03:06:40 -07:00
1 parent d2aa9cf08a
commit 956f53b278
4 files changed
+35 -10
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

vllm_omni/diffusion/diffusion_engine.py
+8
View File
@@ -196,6 +196,10 @@ class DiffusionEngine:
prompt=prompt,
metrics=metrics,
latents=output.trajectory_latents,
trajectory_latents=output.trajectory_latents,
trajectory_timesteps=output.trajectory_timesteps,
trajectory_log_probs=output.trajectory_log_probs,
trajectory_decoded=output.trajectory_decoded,
multimodal_output={"audio": request_audio_payload},
final_output_type="audio",
stage_durations=output.stage_durations,
@@ -252,6 +256,10 @@ class DiffusionEngine:
prompt=prompt,
metrics=metrics,
latents=output.trajectory_latents,
trajectory_latents=output.trajectory_latents,
trajectory_timesteps=output.trajectory_timesteps,
trajectory_log_probs=output.trajectory_log_probs,
trajectory_decoded=output.trajectory_decoded,
multimodal_output={"audio": request_audio_payload},
final_output_type="audio",
stage_durations=output.stage_durations,