| { |
| "title": "Verified Qwen3-Omni LoRA 128-Episode Held-Out Result", |
| "status": "verified_full_128_episode_diagnostic_result", |
| "status_date": "2026-06-08", |
| "backbone": "Qwen/Qwen3-Omni-30B-A3B-Instruct", |
| "adapter": "Qwen3-Omni LoRA", |
| "dataset": "Ropedia Xperience-10M selected 128-episode pilot", |
| "split_policy": { |
| "unit": "episode", |
| "selected_episode_counts": { |
| "test": 16, |
| "train": 96, |
| "val": 16 |
| }, |
| "exported_window_counts": { |
| "train": 2848, |
| "val": 512, |
| "test": 448 |
| }, |
| "exported_episode_counts": { |
| "train": 89, |
| "val": 16, |
| "test": 14 |
| }, |
| "skipped_selected_episodes": 9, |
| "leakage_policy": "Train, validation, and test are separated by episode/session; test windows are used only for held-out evaluation." |
| }, |
| "training": { |
| "num_processes": 8, |
| "epochs": 4, |
| "lora_rank": 16, |
| "lora_alpha": 32, |
| "lora_dropout": 0.05, |
| "num_train_samples": 2848, |
| "num_val_samples": 512, |
| "history": [ |
| { |
| "epoch": 1, |
| "train_loss": 0.40796751019628613, |
| "val_loss": 0.03258896619081497, |
| "global_step": 356 |
| }, |
| { |
| "epoch": 2, |
| "train_loss": 0.027628723937453012, |
| "val_loss": 0.027754632756114006, |
| "global_step": 712 |
| }, |
| { |
| "epoch": 3, |
| "train_loss": 0.02446955946807781, |
| "val_loss": 0.026343274861574173, |
| "global_step": 1068 |
| }, |
| { |
| "epoch": 4, |
| "train_loss": 0.022728607045444712, |
| "val_loss": 0.025629229843616486, |
| "global_step": 1424 |
| } |
| ], |
| "loss": "answer-token cross entropy over supervised JSON tokens", |
| "note": "This current Qwen3-Omni LoRA result is the v4 four-epoch full held-out evaluation on the selected 96/16/16 episode setup." |
| }, |
| "evaluation": { |
| "split": "test", |
| "num_samples": 448, |
| "held_out_episode_count": 14, |
| "json_validity_rate": 1.0, |
| "action_macro_f1": 0.0018678269676001454, |
| "subtask_accuracy": 0.0, |
| "transition_accuracy": 0.9732142857142857, |
| "next_action_accuracy": 0.033482142857142856, |
| "contact_accuracy": 0.7299107142857143, |
| "object_micro_f1": 0.31099781500364165, |
| "quality_target": { |
| "json_validity_rate": 0.98, |
| "status": "met" |
| }, |
| "previous_strict_label_v3_action_macro_f1": 0.0021983997167007384, |
| "previous_structured_json_v2_json_validity_rate": 0.9977678571428571 |
| }, |
| "interpretation": "This is the current verified Qwen3-Omni LoRA diagnostic result for the selected 128-episode setup. The v4 four-epoch package reaches 100% JSON validity and slightly improves next-action, contact, and object metrics versus the prior strict-label v3 package, while action and subtask classification remain weak on held-out episodes. Treat it as a diagnostic baseline and error-analysis source, not as a strong Xperience-10M action recognizer.", |
| "public_package": { |
| "path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full", |
| "audit_status": "pass", |
| "contains_raw_xperience10m_data": false, |
| "contains_qwen_base_weights": false, |
| "contains_lora_weights": false, |
| "adapter_weights_repo": "cy0307/ropedia-qwen3-omni-lora-128ep" |
| }, |
| "required_next_steps": [ |
| "Use the v4 predictions for action/subtask error analysis, unseen-label debugging, and hierarchical action-family scoring.", |
| "Use TASK_SUITE_ENHANCEMENT_128.md and docs/data/task_suite_enhancement_128.json for the no-new-episode suite push before requesting more storage.", |
| "Keep the existing Qwen LoRA adapter repository as the weight-bearing artifact and publish future Qwen v5 runs as separate verified packages.", |
| "Use the verified Cosmos3-Super Forward-Dynamics LoRA package as a separate world-model branch: it updates adapter weights over camera-pose proxy future-vision-velocity targets, not Qwen-style JSON action labels." |
| ] |
| } |
|
|
