nodes/hao-ai-lab-FastVideo
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[refactor] use existing workflow namespace for InterleaveThinker

Browse Source
This commit is contained in:
Mac Lee
2026-06-21 20:37:13 +00:00
parent 58256b1282
commit bb1e8935ee
21 changed files with 66 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/design/interleave_thinker.md
+4 -4
View File
@@ -44,7 +44,7 @@ Training examples live under:
### Orchestration Helpers
The Python helper layer under `fastvideo.workflows.interleave_thinker` is intentionally
The Python helper layer under `fastvideo.workflow.interleave_thinker` is intentionally
not registered as a CLI or server contract. It provides reusable dataclasses,
provider adapters, image-backend adapters, trace serialization, prompt-set
execution helpers, and saved-trace metrics for tests, examples, and downstream
@@ -61,9 +61,9 @@ The runnable example is:
| Planner and critic actors | `fastvideo/train/models/interleave_thinker/` | Wrap Transformers Qwen3-VL checkpoints. They are training actors, not diffusion pipeline components. |
| RL/SFT algorithms | `fastvideo/train/methods/` | Own loss, reward aggregation, advantage computation, KL, and optimizer cadence. |
| Rewards and API clients | `fastvideo/train/methods/rl/rewards/` | Offline reward aggregation is separate from network-backed Gemini/Nano Banana clients. |
| Image generation/editing helpers | `fastvideo/workflows/interleave_thinker/generator.py` | Presents a small image backend protocol for FastVideo, Nano Banana, and fake backends. |
| Runtime orchestration helpers | `fastvideo/workflows/interleave_thinker/` | Plans steps, calls generator/edit backends, calls critic providers, records traces. |
| Evaluation helpers | `fastvideo/workflows/interleave_thinker/evaluation.py` and `trace_eval.py` | Prompt-set execution and saved-trace reporting remain outside training methods. |
| Image generation/editing helpers | `fastvideo/workflow/interleave_thinker/generator.py` | Presents a small image backend protocol for FastVideo, Nano Banana, and fake backends. |
| Runtime orchestration helpers | `fastvideo/workflow/interleave_thinker/` | Plans steps, calls generator/edit backends, calls critic providers, records traces. |
| Evaluation helpers | `fastvideo/workflow/interleave_thinker/evaluation.py` and `trace_eval.py` | Prompt-set execution and saved-trace reporting remain outside training methods. |
This keeps the Qwen actor implementation reusable by SFT, planner GRPO, critic
GRPO, and inference providers without coupling those paths to a specific