Add files using upload-large-folder tool

QWEN3_OMNI_RUN_LINEAGE.md

@@ -0,0 +1,36 @@
+# Qwen3-Omni v1-v6 Run Lineage
+
+Generated: `2026-06-21T09:58:19+00:00`.
+
+Scope: Verified public-safe Qwen3-Omni LoRA/eval packages over the selected Xperience-10M 128-episode surface.
+
+Interpretation rule: Do not confuse the Qwen run versions with the project-level public result layers. The 20-task matrix uses Qwen3-Omni v6 LoRA; v5 remains the pinned prior release; v1-v4 are lineage and ablation evidence.
+
+## Compact Lineage
+
+| Version | What changed | Eval samples | JSON validity | Action macro-F1 | Subtask acc. | Contact acc. | Public role |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| v1 | Selected-128 validation-aware LoRA baseline | 448 | 0.8750 | 0.0027 | 0.0067 | 0.6451 | superseded lineage evidence, not the current 20-task Qwen row |
+| v2 | Structured-JSON reuse full-8-GPU LoRA | 448 | 0.9978 | 0.0024 | 0.0022 | 0.7188 | superseded lineage evidence, not the current 20-task Qwen row |
+| v3 | Strict-label prompt evaluation | 448 | 1.0000 | 0.0022 | 0.0022 | 0.7210 | superseded prompt/eval lineage evidence |
+| v4 | Four-epoch structured-JSON LoRA | 448 | 1.0000 | 0.0019 | 0.0000 | 0.7299 | superseded lineage evidence, not the current 20-task Qwen row |
+| v5 | Multiscale cap96 LoRA | 4032 | 1.0000 | 0.0023 | 0.0112 | 0.7865 | pinned prior release row and comparison baseline |
+| v6 | Rank64 lr5e-5 multiscale LoRA | 4032 | 0.9990 | 0.0029 | 0.0037 | 0.8177 | current public 20-task Qwen3-Omni v6 LoRA row |
+
+## Run IDs And Packages
+
+| Version | Train run | Eval run | Role | Package |
+| --- | --- | --- | --- | --- |
+| v1 | xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_lora | xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval | First verified 96/16/16 selected-episode Qwen3-Omni LoRA package; establishes dataset, training, eval, and packaging plumbing. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval |
+| v2 | xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora | xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full | Reuses the selected-128 split with a stricter structured JSON answer contract and full 8-GPU LoRA training. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full |
+| v3 | xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora | xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full | Strict-label prompt/eval pass over the v2 adapter; improves JSON validity without introducing a new adapter training run. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full |
+| v4 | xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora | xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full | Four-epoch full-8-GPU LoRA run on the same selected split; useful for overfit/metric tradeoff analysis. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full |
+| v5 | xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora | xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full | Dense/multiscale selected-128 run with 4,032 held-out predictions; kept as the pinned prior release because several metrics remain stronger than v6. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full |
+| v6 | xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora | xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full | Current verified Qwen3-Omni row: rank64/lr5e-5 multiscale LoRA plus task-specific probe artifacts used for the 20/20 Qwen matrix coverage. | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full |
+
+## Related Engineering Artifacts
+
+| Artifact | Path | Role |
+| --- | --- | --- |
+| Full-parameter gates | results/omni_finetune/QWEN3_FULL_PARAMETER_GATES_20260609.md | Feasibility and short-train gates; not a public 20-task matrix method row. |
+| Alternate fullsplit v6 package | results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_fullsplit_fast8gpu_lora_fsdp_full_train_noval_tail_logits_fullstatesave_v6_eval_test_full | Verified alternate no-validation/fullsplit artifact retained for audit, not the current matrix row. |

data/mirror_parity.json

@@ -1,9 +1,9 @@
 {
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:09:43+00:00",
+  "generated_at_utc": "2026-06-21T10:25:58+00:00",
   "hf_root": "hf_publish",
   "summary": {
-    "group_count": 1253,
+    "group_count": 1256,
     "failure_count": 0,
     "failures_by_surface": {}
   },
@@ -1063,6 +1063,55 @@
       },
       "failures": []
     },
+    {
+      "name": "data/qwen3_omni_run_lineage.json",
+      "status": "pass",
+      "local": {
+        "path": "repo:docs/data/qwen3_omni_run_lineage.json",
+        "exists": true,
+        "bytes": 9211,
+        "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts_data": {
+          "path": "hf_artifacts:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_data": {
+          "path": "hf_model:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_docs_data": {
+          "path": "hf_model:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model": {
+          "path": "hf_model:metrics/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "data/qwen3_v5_v6_comparison.json",
       "status": "pass",
@@ -1119,44 +1168,44 @@
         "path": "repo:docs/data/quality_gates.json",
         "exists": true,
         "bytes": 8640,
-        "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+        "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts_data": {
           "path": "hf_artifacts:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_data": {
           "path": "hf_model:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_docs_data": {
           "path": "hf_model:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model": {
           "path": "hf_model:metrics/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         }
       },
       "failures": []
@@ -5348,6 +5397,31 @@
       },
       "failures": []
     },
+    {
+      "name": "scripts/build_qwen3_omni_run_lineage.py",
+      "status": "pass",
+      "local": {
+        "path": "repo:scripts/build_qwen3_omni_run_lineage.py",
+        "exists": true,
+        "bytes": 7934,
+        "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+      },
+      "mirrors": {
+        "hf_artifacts": {
+          "path": "hf_artifacts:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        },
+        "hf_model": {
+          "path": "hf_model:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "scripts/build_public_surface_qa.py",
       "status": "pass",
@@ -5829,21 +5903,21 @@
       "local": {
         "path": "repo:scripts/validate_mirror_parity.py",
         "exists": true,
-        "bytes": 36363,
-        "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+        "bytes": 36470,
+        "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         },
         "hf_model": {
           "path": "hf_model:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         }
       },
       "failures": []
@@ -6004,21 +6078,21 @@
       "local": {
         "path": "repo:scripts/sync_hf_publish_mirrors.py",
         "exists": true,
-        "bytes": 28655,
-        "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+        "bytes": 29242,
+        "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         },
         "hf_model": {
           "path": "hf_model:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         }
       },
       "failures": []
@@ -30843,26 +30917,26 @@
         "path": "repo:QUALITY_GATES.md",
         "exists": true,
         "bytes": 5184,
-        "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+        "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_model": {
           "path": "hf_model:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         }
       },
       "failures": []
@@ -31084,6 +31158,37 @@
       },
       "failures": []
     },
+    {
+      "name": "docs/QWEN3_OMNI_RUN_LINEAGE.md",
+      "status": "pass",
+      "local": {
+        "path": "repo:QWEN3_OMNI_RUN_LINEAGE.md",
+        "exists": true,
+        "bytes": 4595,
+        "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_model": {
+          "path": "hf_model:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "docs/PROJECT_BRIEF.md",
       "status": "pass",

data/quality_gates.json

@@ -1,7 +1,7 @@
 {
   "title": "Ropedia Xperience-10M Release Checks",
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:10:44+00:00",
+  "generated_at_utc": "2026-06-21T10:25:40+00:00",
   "rule": "A release is current when the automated reports pass and the live GitHub/Hugging Face mirrors are verified after publishing.",
   "automated_gates": [
     {

data/qwen3_omni_run_lineage.json

@@ -0,0 +1,161 @@
+{
+  "current_public_matrix_row": "qwen3_omni_v6_lora",
+  "generated_at_utc": "2026-06-21T09:58:19+00:00",
+  "interpretation_rule": "Do not confuse the Qwen run versions with the project-level public result layers. The 20-task matrix uses Qwen3-Omni v6 LoRA; v5 remains the pinned prior release; v1-v4 are lineage and ablation evidence.",
+  "pinned_prior_release": "v5",
+  "related_engineering_artifacts": [
+    {
+      "name": "Full-parameter gates",
+      "path": "results/omni_finetune/QWEN3_FULL_PARAMETER_GATES_20260609.md",
+      "role": "Feasibility and short-train gates; not a public 20-task matrix method row."
+    },
+    {
+      "name": "Alternate fullsplit v6 package",
+      "path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_fullsplit_fast8gpu_lora_fsdp_full_train_noval_tail_logits_fullstatesave_v6_eval_test_full",
+      "role": "Verified alternate no-validation/fullsplit artifact retained for audit, not the current matrix row."
+    }
+  ],
+  "runs": [
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0026621494447581404,
+        "contact_accuracy": 0.6450892857142857,
+        "json_validity_rate": 0.875,
+        "next_action_accuracy": 0.024553571428571428,
+        "object_micro_f1": 0.22299431459254582,
+        "subtask_accuracy": 0.006696428571428571,
+        "transition_accuracy": 0.8504464285714286
+      },
+      "package": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "First verified 96/16/16 selected-episode Qwen3-Omni LoRA package; establishes dataset, training, eval, and packaging plumbing.",
+      "status": "verified",
+      "title": "Selected-128 validation-aware LoRA baseline",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_lora",
+      "version": "v1"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0024331644885523347,
+        "contact_accuracy": 0.71875,
+        "json_validity_rate": 0.9977678571428571,
+        "next_action_accuracy": 0.029017857142857144,
+        "object_micro_f1": 0.30160427807486634,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9709821428571429
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Reuses the selected-128 split with a stricter structured JSON answer contract and full 8-GPU LoRA training.",
+      "status": "verified",
+      "title": "Structured-JSON reuse full-8-GPU LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v2"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0021983997167007384,
+        "contact_accuracy": 0.7209821428571429,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.03125,
+        "object_micro_f1": 0.30688228657389993,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "public_matrix_role": "superseded prompt/eval lineage evidence",
+      "role": "Strict-label prompt/eval pass over the v2 adapter; improves JSON validity without introducing a new adapter training run.",
+      "status": "verified",
+      "title": "Strict-label prompt evaluation",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v3"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0018678269676001454,
+        "contact_accuracy": 0.7299107142857143,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.033482142857142856,
+        "object_micro_f1": 0.31099781500364165,
+        "subtask_accuracy": 0.0,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Four-epoch full-8-GPU LoRA run on the same selected split; useful for overfit/metric tradeoff analysis.",
+      "status": "verified",
+      "title": "Four-epoch structured-JSON LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora",
+      "version": "v4"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.002289711036077459,
+        "contact_accuracy": 0.7864583333333334,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.053618594823032224,
+        "object_micro_f1": 0.31614599936244814,
+        "subtask_accuracy": 0.011194029850746268,
+        "transition_accuracy": 0.9908234126984127
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "pinned prior release row and comparison baseline",
+      "role": "Dense/multiscale selected-128 run with 4,032 held-out predictions; kept as the pinned prior release because several metrics remain stronger than v6.",
+      "status": "verified",
+      "title": "Multiscale cap96 LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "version": "v5"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.0028830723979596335,
+        "contact_accuracy": 0.8177083333333334,
+        "json_validity_rate": 0.9990079365079365,
+        "next_action_accuracy": 0.04305335446381405,
+        "object_micro_f1": 0.3064982378331287,
+        "subtask_accuracy": 0.0037313432835820895,
+        "transition_accuracy": 0.9898313492063492
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "current public 20-task Qwen3-Omni v6 LoRA row",
+      "role": "Current verified Qwen3-Omni row: rank64/lr5e-5 multiscale LoRA plus task-specific probe artifacts used for the 20/20 Qwen matrix coverage.",
+      "status": "verified",
+      "title": "Rank64 lr5e-5 multiscale LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora",
+      "version": "v6"
+    }
+  ],
+  "scope": "Verified public-safe Qwen3-Omni LoRA/eval packages over the selected Xperience-10M 128-episode surface.",
+  "status": "pass",
+  "title": "Qwen3-Omni v1-v6 Run Lineage"
+}

docs/data/mirror_parity.json

@@ -1,9 +1,9 @@
 {
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:09:43+00:00",
+  "generated_at_utc": "2026-06-21T10:25:58+00:00",
   "hf_root": "hf_publish",
   "summary": {
-    "group_count": 1253,
+    "group_count": 1256,
     "failure_count": 0,
     "failures_by_surface": {}
   },
@@ -1063,6 +1063,55 @@
       },
       "failures": []
     },
+    {
+      "name": "data/qwen3_omni_run_lineage.json",
+      "status": "pass",
+      "local": {
+        "path": "repo:docs/data/qwen3_omni_run_lineage.json",
+        "exists": true,
+        "bytes": 9211,
+        "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts_data": {
+          "path": "hf_artifacts:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_data": {
+          "path": "hf_model:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_docs_data": {
+          "path": "hf_model:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model": {
+          "path": "hf_model:metrics/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "data/qwen3_v5_v6_comparison.json",
       "status": "pass",
@@ -1119,44 +1168,44 @@
         "path": "repo:docs/data/quality_gates.json",
         "exists": true,
         "bytes": 8640,
-        "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+        "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts_data": {
           "path": "hf_artifacts:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_data": {
           "path": "hf_model:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_docs_data": {
           "path": "hf_model:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model": {
           "path": "hf_model:metrics/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         }
       },
       "failures": []
@@ -5348,6 +5397,31 @@
       },
       "failures": []
     },
+    {
+      "name": "scripts/build_qwen3_omni_run_lineage.py",
+      "status": "pass",
+      "local": {
+        "path": "repo:scripts/build_qwen3_omni_run_lineage.py",
+        "exists": true,
+        "bytes": 7934,
+        "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+      },
+      "mirrors": {
+        "hf_artifacts": {
+          "path": "hf_artifacts:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        },
+        "hf_model": {
+          "path": "hf_model:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "scripts/build_public_surface_qa.py",
       "status": "pass",
@@ -5829,21 +5903,21 @@
       "local": {
         "path": "repo:scripts/validate_mirror_parity.py",
         "exists": true,
-        "bytes": 36363,
-        "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+        "bytes": 36470,
+        "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         },
         "hf_model": {
           "path": "hf_model:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         }
       },
       "failures": []
@@ -6004,21 +6078,21 @@
       "local": {
         "path": "repo:scripts/sync_hf_publish_mirrors.py",
         "exists": true,
-        "bytes": 28655,
-        "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+        "bytes": 29242,
+        "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         },
         "hf_model": {
           "path": "hf_model:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         }
       },
       "failures": []
@@ -30843,26 +30917,26 @@
         "path": "repo:QUALITY_GATES.md",
         "exists": true,
         "bytes": 5184,
-        "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+        "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_model": {
           "path": "hf_model:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         }
       },
       "failures": []
@@ -31084,6 +31158,37 @@
       },
       "failures": []
     },
+    {
+      "name": "docs/QWEN3_OMNI_RUN_LINEAGE.md",
+      "status": "pass",
+      "local": {
+        "path": "repo:QWEN3_OMNI_RUN_LINEAGE.md",
+        "exists": true,
+        "bytes": 4595,
+        "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_model": {
+          "path": "hf_model:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "docs/PROJECT_BRIEF.md",
       "status": "pass",

docs/data/quality_gates.json

@@ -1,7 +1,7 @@
 {
   "title": "Ropedia Xperience-10M Release Checks",
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:10:44+00:00",
+  "generated_at_utc": "2026-06-21T10:25:40+00:00",
   "rule": "A release is current when the automated reports pass and the live GitHub/Hugging Face mirrors are verified after publishing.",
   "automated_gates": [
     {

docs/data/qwen3_omni_run_lineage.json

@@ -0,0 +1,161 @@
+{
+  "current_public_matrix_row": "qwen3_omni_v6_lora",
+  "generated_at_utc": "2026-06-21T09:58:19+00:00",
+  "interpretation_rule": "Do not confuse the Qwen run versions with the project-level public result layers. The 20-task matrix uses Qwen3-Omni v6 LoRA; v5 remains the pinned prior release; v1-v4 are lineage and ablation evidence.",
+  "pinned_prior_release": "v5",
+  "related_engineering_artifacts": [
+    {
+      "name": "Full-parameter gates",
+      "path": "results/omni_finetune/QWEN3_FULL_PARAMETER_GATES_20260609.md",
+      "role": "Feasibility and short-train gates; not a public 20-task matrix method row."
+    },
+    {
+      "name": "Alternate fullsplit v6 package",
+      "path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_fullsplit_fast8gpu_lora_fsdp_full_train_noval_tail_logits_fullstatesave_v6_eval_test_full",
+      "role": "Verified alternate no-validation/fullsplit artifact retained for audit, not the current matrix row."
+    }
+  ],
+  "runs": [
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0026621494447581404,
+        "contact_accuracy": 0.6450892857142857,
+        "json_validity_rate": 0.875,
+        "next_action_accuracy": 0.024553571428571428,
+        "object_micro_f1": 0.22299431459254582,
+        "subtask_accuracy": 0.006696428571428571,
+        "transition_accuracy": 0.8504464285714286
+      },
+      "package": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "First verified 96/16/16 selected-episode Qwen3-Omni LoRA package; establishes dataset, training, eval, and packaging plumbing.",
+      "status": "verified",
+      "title": "Selected-128 validation-aware LoRA baseline",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_lora",
+      "version": "v1"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0024331644885523347,
+        "contact_accuracy": 0.71875,
+        "json_validity_rate": 0.9977678571428571,
+        "next_action_accuracy": 0.029017857142857144,
+        "object_micro_f1": 0.30160427807486634,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9709821428571429
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Reuses the selected-128 split with a stricter structured JSON answer contract and full 8-GPU LoRA training.",
+      "status": "verified",
+      "title": "Structured-JSON reuse full-8-GPU LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v2"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0021983997167007384,
+        "contact_accuracy": 0.7209821428571429,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.03125,
+        "object_micro_f1": 0.30688228657389993,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "public_matrix_role": "superseded prompt/eval lineage evidence",
+      "role": "Strict-label prompt/eval pass over the v2 adapter; improves JSON validity without introducing a new adapter training run.",
+      "status": "verified",
+      "title": "Strict-label prompt evaluation",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v3"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0018678269676001454,
+        "contact_accuracy": 0.7299107142857143,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.033482142857142856,
+        "object_micro_f1": 0.31099781500364165,
+        "subtask_accuracy": 0.0,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Four-epoch full-8-GPU LoRA run on the same selected split; useful for overfit/metric tradeoff analysis.",
+      "status": "verified",
+      "title": "Four-epoch structured-JSON LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora",
+      "version": "v4"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.002289711036077459,
+        "contact_accuracy": 0.7864583333333334,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.053618594823032224,
+        "object_micro_f1": 0.31614599936244814,
+        "subtask_accuracy": 0.011194029850746268,
+        "transition_accuracy": 0.9908234126984127
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "pinned prior release row and comparison baseline",
+      "role": "Dense/multiscale selected-128 run with 4,032 held-out predictions; kept as the pinned prior release because several metrics remain stronger than v6.",
+      "status": "verified",
+      "title": "Multiscale cap96 LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "version": "v5"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.0028830723979596335,
+        "contact_accuracy": 0.8177083333333334,
+        "json_validity_rate": 0.9990079365079365,
+        "next_action_accuracy": 0.04305335446381405,
+        "object_micro_f1": 0.3064982378331287,
+        "subtask_accuracy": 0.0037313432835820895,
+        "transition_accuracy": 0.9898313492063492
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "current public 20-task Qwen3-Omni v6 LoRA row",
+      "role": "Current verified Qwen3-Omni row: rank64/lr5e-5 multiscale LoRA plus task-specific probe artifacts used for the 20/20 Qwen matrix coverage.",
+      "status": "verified",
+      "title": "Rank64 lr5e-5 multiscale LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora",
+      "version": "v6"
+    }
+  ],
+  "scope": "Verified public-safe Qwen3-Omni LoRA/eval packages over the selected Xperience-10M 128-episode surface.",
+  "status": "pass",
+  "title": "Qwen3-Omni v1-v6 Run Lineage"
+}

metrics/mirror_parity.json

@@ -1,9 +1,9 @@
 {
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:09:43+00:00",
+  "generated_at_utc": "2026-06-21T10:25:58+00:00",
   "hf_root": "hf_publish",
   "summary": {
-    "group_count": 1253,
+    "group_count": 1256,
     "failure_count": 0,
     "failures_by_surface": {}
   },
@@ -1063,6 +1063,55 @@
       },
       "failures": []
     },
+    {
+      "name": "data/qwen3_omni_run_lineage.json",
+      "status": "pass",
+      "local": {
+        "path": "repo:docs/data/qwen3_omni_run_lineage.json",
+        "exists": true,
+        "bytes": 9211,
+        "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts_data": {
+          "path": "hf_artifacts:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_data": {
+          "path": "hf_model:data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model_docs_data": {
+          "path": "hf_model:docs/data/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        },
+        "hf_model": {
+          "path": "hf_model:metrics/qwen3_omni_run_lineage.json",
+          "exists": true,
+          "bytes": 9211,
+          "sha256": "98e322fa0bec38aab93b924f4f1fcd13eee848bf9b6791fea7adcfb0db3c2953"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "data/qwen3_v5_v6_comparison.json",
       "status": "pass",
@@ -1119,44 +1168,44 @@
         "path": "repo:docs/data/quality_gates.json",
         "exists": true,
         "bytes": 8640,
-        "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+        "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts_data": {
           "path": "hf_artifacts:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_data": {
           "path": "hf_model:data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model_docs_data": {
           "path": "hf_model:docs/data/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         },
         "hf_model": {
           "path": "hf_model:metrics/quality_gates.json",
           "exists": true,
           "bytes": 8640,
-          "sha256": "6a2890d168d4f09510a4c59cdf9319e6e03d1e5ef2a50ac57bff644565673bf9"
+          "sha256": "3cb0aca2dca01448cb9bc5cbb519a91bc6397c08a1eaaa84c031e773221e5a0a"
         }
       },
       "failures": []
@@ -5348,6 +5397,31 @@
       },
       "failures": []
     },
+    {
+      "name": "scripts/build_qwen3_omni_run_lineage.py",
+      "status": "pass",
+      "local": {
+        "path": "repo:scripts/build_qwen3_omni_run_lineage.py",
+        "exists": true,
+        "bytes": 7934,
+        "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+      },
+      "mirrors": {
+        "hf_artifacts": {
+          "path": "hf_artifacts:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        },
+        "hf_model": {
+          "path": "hf_model:scripts/build_qwen3_omni_run_lineage.py",
+          "exists": true,
+          "bytes": 7934,
+          "sha256": "76b68613e3470d3a00642a516e9fb1272de94503d22877c845af105e8d651a6e"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "scripts/build_public_surface_qa.py",
       "status": "pass",
@@ -5829,21 +5903,21 @@
       "local": {
         "path": "repo:scripts/validate_mirror_parity.py",
         "exists": true,
-        "bytes": 36363,
-        "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+        "bytes": 36470,
+        "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         },
         "hf_model": {
           "path": "hf_model:scripts/validate_mirror_parity.py",
           "exists": true,
-          "bytes": 36363,
-          "sha256": "24cfad669a2f77794e30acf75c0df5560991f2aab110d751f73e4f0739b2b06f"
+          "bytes": 36470,
+          "sha256": "0617ffc2fd6ed22b4202c4ec532e50796923e74b4ccd7e1fb3e1ff301995bb2c"
         }
       },
       "failures": []
@@ -6004,21 +6078,21 @@
       "local": {
         "path": "repo:scripts/sync_hf_publish_mirrors.py",
         "exists": true,
-        "bytes": 28655,
-        "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+        "bytes": 29242,
+        "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
       },
       "mirrors": {
         "hf_artifacts": {
           "path": "hf_artifacts:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         },
         "hf_model": {
           "path": "hf_model:scripts/sync_hf_publish_mirrors.py",
           "exists": true,
-          "bytes": 28655,
-          "sha256": "75ec8d1c324317c8f8e5c4a483e7aef43e95c30ee3b581c0108d1de059cd68f0"
+          "bytes": 29242,
+          "sha256": "60566f5c6922d02c249a33d6c57b9474453369de038a30c9285febc7bd0347fa"
         }
       },
       "failures": []
@@ -30843,26 +30917,26 @@
         "path": "repo:QUALITY_GATES.md",
         "exists": true,
         "bytes": 5184,
-        "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+        "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
       },
       "mirrors": {
         "hf_space": {
           "path": "hf_space:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_artifacts": {
           "path": "hf_artifacts:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         },
         "hf_model": {
           "path": "hf_model:QUALITY_GATES.md",
           "exists": true,
           "bytes": 5184,
-          "sha256": "01bc15b23784df0668624b819ead739058628378f38435f41655a1491b1e8465"
+          "sha256": "4931d4457c4c5b0978fdf31861b6e3e2da6e24368398cf1756120a32cbff98f0"
         }
       },
       "failures": []
@@ -31084,6 +31158,37 @@
       },
       "failures": []
     },
+    {
+      "name": "docs/QWEN3_OMNI_RUN_LINEAGE.md",
+      "status": "pass",
+      "local": {
+        "path": "repo:QWEN3_OMNI_RUN_LINEAGE.md",
+        "exists": true,
+        "bytes": 4595,
+        "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+      },
+      "mirrors": {
+        "hf_space": {
+          "path": "hf_space:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_artifacts": {
+          "path": "hf_artifacts:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        },
+        "hf_model": {
+          "path": "hf_model:QWEN3_OMNI_RUN_LINEAGE.md",
+          "exists": true,
+          "bytes": 4595,
+          "sha256": "dfb1b8e931cc2dd7a15e8e3489047b5ee198495932961680b86875c169f92a68"
+        }
+      },
+      "failures": []
+    },
     {
       "name": "docs/PROJECT_BRIEF.md",
       "status": "pass",

metrics/quality_gates.json

@@ -1,7 +1,7 @@
 {
   "title": "Ropedia Xperience-10M Release Checks",
   "status": "pass",
-  "generated_at_utc": "2026-06-21T10:10:44+00:00",
+  "generated_at_utc": "2026-06-21T10:25:40+00:00",
   "rule": "A release is current when the automated reports pass and the live GitHub/Hugging Face mirrors are verified after publishing.",
   "automated_gates": [
     {

metrics/qwen3_omni_run_lineage.json

@@ -0,0 +1,161 @@
+{
+  "current_public_matrix_row": "qwen3_omni_v6_lora",
+  "generated_at_utc": "2026-06-21T09:58:19+00:00",
+  "interpretation_rule": "Do not confuse the Qwen run versions with the project-level public result layers. The 20-task matrix uses Qwen3-Omni v6 LoRA; v5 remains the pinned prior release; v1-v4 are lineage and ablation evidence.",
+  "pinned_prior_release": "v5",
+  "related_engineering_artifacts": [
+    {
+      "name": "Full-parameter gates",
+      "path": "results/omni_finetune/QWEN3_FULL_PARAMETER_GATES_20260609.md",
+      "role": "Feasibility and short-train gates; not a public 20-task matrix method row."
+    },
+    {
+      "name": "Alternate fullsplit v6 package",
+      "path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_fullsplit_fast8gpu_lora_fsdp_full_train_noval_tail_logits_fullstatesave_v6_eval_test_full",
+      "role": "Verified alternate no-validation/fullsplit artifact retained for audit, not the current matrix row."
+    }
+  ],
+  "runs": [
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0026621494447581404,
+        "contact_accuracy": 0.6450892857142857,
+        "json_validity_rate": 0.875,
+        "next_action_accuracy": 0.024553571428571428,
+        "object_micro_f1": 0.22299431459254582,
+        "subtask_accuracy": 0.006696428571428571,
+        "transition_accuracy": 0.8504464285714286
+      },
+      "package": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "First verified 96/16/16 selected-episode Qwen3-Omni LoRA package; establishes dataset, training, eval, and packaging plumbing.",
+      "status": "verified",
+      "title": "Selected-128 validation-aware LoRA baseline",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_lora",
+      "version": "v1"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0024331644885523347,
+        "contact_accuracy": 0.71875,
+        "json_validity_rate": 0.9977678571428571,
+        "next_action_accuracy": 0.029017857142857144,
+        "object_micro_f1": 0.30160427807486634,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9709821428571429
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Reuses the selected-128 split with a stricter structured JSON answer contract and full 8-GPU LoRA training.",
+      "status": "verified",
+      "title": "Structured-JSON reuse full-8-GPU LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v2"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0021983997167007384,
+        "contact_accuracy": 0.7209821428571429,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.03125,
+        "object_micro_f1": 0.30688228657389993,
+        "subtask_accuracy": 0.002232142857142857,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+      "public_matrix_role": "superseded prompt/eval lineage evidence",
+      "role": "Strict-label prompt/eval pass over the v2 adapter; improves JSON validity without introducing a new adapter training run.",
+      "status": "verified",
+      "title": "Strict-label prompt evaluation",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora",
+      "version": "v3"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "eval_samples": 448,
+      "metrics": {
+        "action_macro_f1": 0.0018678269676001454,
+        "contact_accuracy": 0.7299107142857143,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.033482142857142856,
+        "object_micro_f1": 0.31099781500364165,
+        "subtask_accuracy": 0.0,
+        "transition_accuracy": 0.9732142857142857
+      },
+      "package": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+      "role": "Four-epoch full-8-GPU LoRA run on the same selected split; useful for overfit/metric tradeoff analysis.",
+      "status": "verified",
+      "title": "Four-epoch structured-JSON LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora",
+      "version": "v4"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.002289711036077459,
+        "contact_accuracy": 0.7864583333333334,
+        "json_validity_rate": 1.0,
+        "next_action_accuracy": 0.053618594823032224,
+        "object_micro_f1": 0.31614599936244814,
+        "subtask_accuracy": 0.011194029850746268,
+        "transition_accuracy": 0.9908234126984127
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "pinned prior release row and comparison baseline",
+      "role": "Dense/multiscale selected-128 run with 4,032 held-out predictions; kept as the pinned prior release because several metrics remain stronger than v6.",
+      "status": "verified",
+      "title": "Multiscale cap96 LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "version": "v5"
+    },
+    {
+      "dataset_contract": "xperience10m_episode_json_qa_v1",
+      "dataset_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora",
+      "eval_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "eval_samples": 4032,
+      "metrics": {
+        "action_macro_f1": 0.0028830723979596335,
+        "contact_accuracy": 0.8177083333333334,
+        "json_validity_rate": 0.9990079365079365,
+        "next_action_accuracy": 0.04305335446381405,
+        "object_micro_f1": 0.3064982378331287,
+        "subtask_accuracy": 0.0037313432835820895,
+        "transition_accuracy": 0.9898313492063492
+      },
+      "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "package_path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+      "public_matrix_role": "current public 20-task Qwen3-Omni v6 LoRA row",
+      "role": "Current verified Qwen3-Omni row: rank64/lr5e-5 multiscale LoRA plus task-specific probe artifacts used for the 20/20 Qwen matrix coverage.",
+      "status": "verified",
+      "title": "Rank64 lr5e-5 multiscale LoRA",
+      "train_run_id": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora",
+      "version": "v6"
+    }
+  ],
+  "scope": "Verified public-safe Qwen3-Omni LoRA/eval packages over the selected Xperience-10M 128-episode surface.",
+  "status": "pass",
+  "title": "Qwen3-Omni v1-v6 Run Lineage"
+}

scripts/build_qwen3_omni_run_lineage.py

@@ -0,0 +1,204 @@
+#!/usr/bin/env python3
+"""Build the public Qwen3-Omni v1-v6 run-lineage summary."""
+
+from __future__ import annotations
+
+import json
+from datetime import datetime, timezone
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+VERIFIED = ROOT / "results/omni_finetune/verified_public"
+OUTPUT_JSON = ROOT / "docs/data/qwen3_omni_run_lineage.json"
+OUTPUT_MD = ROOT / "QWEN3_OMNI_RUN_LINEAGE.md"
+
+RUNS = [
+    {
+        "version": "v1",
+        "title": "Selected-128 validation-aware LoRA baseline",
+        "package": "xperience10m_qwen3_omni_128ep_96train_16val_16test_valmon_20260605_eval",
+        "role": "First verified 96/16/16 selected-episode Qwen3-Omni LoRA package; establishes dataset, training, eval, and packaging plumbing.",
+        "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+    },
+    {
+        "version": "v2",
+        "title": "Structured-JSON reuse full-8-GPU LoRA",
+        "package": "xperience10m_qwen3_omni_128ep_structured_json_v2_reuse_full8gpu_lora_eval_test_full",
+        "role": "Reuses the selected-128 split with a stricter structured JSON answer contract and full 8-GPU LoRA training.",
+        "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+    },
+    {
+        "version": "v3",
+        "title": "Strict-label prompt evaluation",
+        "package": "xperience10m_qwen3_omni_128ep_structured_json_v3_strict_label_prompt_reuse_lora_eval_test_full",
+        "role": "Strict-label prompt/eval pass over the v2 adapter; improves JSON validity without introducing a new adapter training run.",
+        "public_matrix_role": "superseded prompt/eval lineage evidence",
+    },
+    {
+        "version": "v4",
+        "title": "Four-epoch structured-JSON LoRA",
+        "package": "xperience10m_qwen3_omni_128ep_structured_json_v4_4epoch_full8gpu_lora_eval_test_full",
+        "role": "Four-epoch full-8-GPU LoRA run on the same selected split; useful for overfit/metric tradeoff analysis.",
+        "public_matrix_role": "superseded lineage evidence, not the current 20-task Qwen row",
+    },
+    {
+        "version": "v5",
+        "title": "Multiscale cap96 LoRA",
+        "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v5_full8gpu_lora_eval_test_full",
+        "role": "Dense/multiscale selected-128 run with 4,032 held-out predictions; kept as the pinned prior release because several metrics remain stronger than v6.",
+        "public_matrix_role": "pinned prior release row and comparison baseline",
+    },
+    {
+        "version": "v6",
+        "title": "Rank64 lr5e-5 multiscale LoRA",
+        "package": "xperience10m_qwen3_omni_128ep_multiscale_cap96_v6_rank64_lr5e5_full8gpu_lora_eval_test_full",
+        "role": "Current verified Qwen3-Omni row: rank64/lr5e-5 multiscale LoRA plus task-specific probe artifacts used for the 20/20 Qwen matrix coverage.",
+        "public_matrix_role": "current public 20-task Qwen3-Omni v6 LoRA row",
+    },
+]
+
+METRIC_KEYS = [
+    "json_validity_rate",
+    "action_macro_f1",
+    "subtask_accuracy",
+    "transition_accuracy",
+    "next_action_accuracy",
+    "contact_accuracy",
+    "object_micro_f1",
+]
+
+
+def read_json(path: Path) -> dict:
+    return json.loads(path.read_text(encoding="utf-8"))
+
+
+def metric_value(metrics: dict, key: str):
+    return metrics.get(key)
+
+
+def fmt(value) -> str:
+    if value is None:
+        return ""
+    if isinstance(value, float):
+        return f"{value:.4f}"
+    return str(value)
+
+
+def markdown_table(headers: list[str], rows: list[list[str]]) -> str:
+    out = [
+        "| " + " | ".join(headers) + " |",
+        "| " + " | ".join("---" for _ in headers) + " |",
+    ]
+    for row in rows:
+        out.append("| " + " | ".join(str(cell).replace("|", "\\|").replace("\n", " ") for cell in row) + " |")
+    return "\n".join(out)
+
+
+def build_payload() -> dict:
+    rows = []
+    for spec in RUNS:
+        package_dir = VERIFIED / spec["package"]
+        summary = read_json(package_dir / "verified_result_summary.json")
+        metrics = read_json(package_dir / "eval/metrics.json")
+        row = {
+            **spec,
+            "status": summary.get("status", "verified"),
+            "package_path": str(package_dir.relative_to(ROOT)),
+            "dataset_run_id": summary.get("dataset_run_id"),
+            "train_run_id": summary.get("train_run_id"),
+            "eval_run_id": summary.get("eval_run_id"),
+            "dataset_contract": summary.get("dataset_contract"),
+            "eval_samples": metrics.get("num_samples") or metrics.get("eval_samples"),
+            "metrics": {key: metric_value(metrics, key) for key in METRIC_KEYS},
+        }
+        rows.append(row)
+    return {
+        "title": "Qwen3-Omni v1-v6 Run Lineage",
+        "status": "pass",
+        "generated_at_utc": datetime.now(timezone.utc).isoformat(timespec="seconds"),
+        "scope": "Verified public-safe Qwen3-Omni LoRA/eval packages over the selected Xperience-10M 128-episode surface.",
+        "interpretation_rule": (
+            "Do not confuse the Qwen run versions with the project-level public result layers. "
+            "The 20-task matrix uses Qwen3-Omni v6 LoRA; v5 remains the pinned prior release; "
+            "v1-v4 are lineage and ablation evidence."
+        ),
+        "current_public_matrix_row": "qwen3_omni_v6_lora",
+        "pinned_prior_release": "v5",
+        "runs": rows,
+        "related_engineering_artifacts": [
+            {
+                "name": "Full-parameter gates",
+                "path": "results/omni_finetune/QWEN3_FULL_PARAMETER_GATES_20260609.md",
+                "role": "Feasibility and short-train gates; not a public 20-task matrix method row.",
+            },
+            {
+                "name": "Alternate fullsplit v6 package",
+                "path": "results/omni_finetune/verified_public/xperience10m_qwen3_omni_128ep_fullsplit_fast8gpu_lora_fsdp_full_train_noval_tail_logits_fullstatesave_v6_eval_test_full",
+                "role": "Verified alternate no-validation/fullsplit artifact retained for audit, not the current matrix row.",
+            },
+        ],
+    }
+
+
+def write_outputs(payload: dict) -> None:
+    OUTPUT_JSON.parent.mkdir(parents=True, exist_ok=True)
+    OUTPUT_JSON.write_text(json.dumps(payload, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+
+    rows = []
+    for run in payload["runs"]:
+        m = run["metrics"]
+        rows.append(
+            [
+                run["version"],
+                run["title"],
+                run["eval_samples"],
+                fmt(m["json_validity_rate"]),
+                fmt(m["action_macro_f1"]),
+                fmt(m["subtask_accuracy"]),
+                fmt(m["contact_accuracy"]),
+                run["public_matrix_role"],
+            ]
+        )
+    detail_rows = [
+        [
+            run["version"],
+            run["train_run_id"],
+            run["eval_run_id"],
+            run["role"],
+            run["package_path"],
+        ]
+        for run in payload["runs"]
+    ]
+    text = f"""# Qwen3-Omni v1-v6 Run Lineage
+
+Generated: `{payload['generated_at_utc']}`.
+
+Scope: {payload['scope']}
+
+Interpretation rule: {payload['interpretation_rule']}
+
+## Compact Lineage
+
+{markdown_table(['Version', 'What changed', 'Eval samples', 'JSON validity', 'Action macro-F1', 'Subtask acc.', 'Contact acc.', 'Public role'], rows)}
+
+## Run IDs And Packages
+
+{markdown_table(['Version', 'Train run', 'Eval run', 'Role', 'Package'], detail_rows)}
+
+## Related Engineering Artifacts
+
+{markdown_table(['Artifact', 'Path', 'Role'], [[row['name'], row['path'], row['role']] for row in payload['related_engineering_artifacts']])}
+"""
+    OUTPUT_MD.write_text(text, encoding="utf-8")
+
+
+def main() -> int:
+    write_outputs(build_payload())
+    print(f"Wrote {OUTPUT_JSON.relative_to(ROOT)}")
+    print(f"Wrote {OUTPUT_MD.relative_to(ROOT)}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

scripts/sync_hf_publish_mirrors.py

@@ -53,7 +53,8 @@ stride, feature manifest, chronological split, and minimal/neural head pattern.
 The historical `tier2_task_suite` path is retained only for stable artifact
 links to tasks 13-20. Results are organized as two evidence lines:
 one public sample episode for task construction and reproducibility, and
-128 selected episodes for same-split baselines plus Qwen3/Cosmos comparison.
+128 selected episodes for same-split baselines plus Qwen3-Omni and Cosmos3
+comparison.
 The line map is published as `docs/assets/charts/two_evidence_line_map.svg`
 with machine-readable summaries in `docs/data/two_evidence_lines.json` and
 `docs/data/two_evidence_line_result_summary.json`.
@@ -120,6 +121,11 @@ QWEN_COMPARISON_ROW = (
     "| Compare Qwen3 v5/v6 diagnostic branches | "
     "`docs/data/qwen3_v5_v6_comparison.json` |"
 )
+QWEN_LINEAGE_MARKER = "docs/data/qwen3_omni_run_lineage.json"
+QWEN_LINEAGE_ROW = (
+    "| Explain Qwen3-Omni v1-v6 run lineage | "
+    "`QWEN3_OMNI_RUN_LINEAGE.md`, `docs/data/qwen3_omni_run_lineage.json` |"
+)
 QWEN_ARTIFACT_OLD_BULLET = """- A current verified Qwen3-Omni strict-label v3 held-out package for the
   selected 96/16/16 episode split, with 100.00% JSON validity and weak
   action/subtask quality documented as the next error-analysis target."""
@@ -460,12 +466,19 @@ def ensure_current_qwen_card_links(hf_root: Path, *, dry_run: bool) -> list[str]
         if QWEN_COMPARISON_ROW not in text:
             anchor = "| Compare current versions and model groups | `docs/data/omni_model_comparison.json` |"
             text = text.replace(anchor, anchor + "\n" + QWEN_COMPARISON_ROW, 1)
+        if QWEN_LINEAGE_ROW not in text:
+            text = text.replace(QWEN_COMPARISON_ROW, QWEN_COMPARISON_ROW + "\n" + QWEN_LINEAGE_ROW, 1)
         if QWEN_ARTIFACT_OLD_BULLET in text:
             text = text.replace(QWEN_ARTIFACT_OLD_BULLET, QWEN_ARTIFACT_CURRENT_BULLET, 1)
         if "99.90% JSON validity" not in text:
             text = text.rstrip() + "\n\n" + QWEN_ARTIFACT_CURRENT_BULLET + "\n"
         if QWEN_COMPARISON_MARKER not in text:
             text = text.rstrip() + f"\n\nQwen v5/v6 comparison: `{QWEN_COMPARISON_MARKER}`.\n"
+        if QWEN_LINEAGE_MARKER not in text:
+            text = text.rstrip() + (
+                f"\n\nQwen3-Omni v1-v6 lineage: `QWEN3_OMNI_RUN_LINEAGE.md` "
+                f"and `{QWEN_LINEAGE_MARKER}`.\n"
+            )
         if text != original:
             updated.append("artifacts/README.md")
             if not dry_run:

scripts/validate_mirror_parity.py

@@ -62,6 +62,7 @@ DATA_FILES = [
     "publication_audit.json",
     "public_surface_qa.json",
     "qwen3_full_parameter_gates.json",
+    "qwen3_omni_run_lineage.json",
     "qwen3_v5_v6_comparison.json",
     "quality_gates.json",
     "raw_sample_files.json",
@@ -189,6 +190,7 @@ SCRIPT_FILES = [
     "build_evaluation_protocol.py",
     "build_figure_index.py",
     "build_quality_gates.py",
+    "build_qwen3_omni_run_lineage.py",
     "build_public_surface_qa.py",
     "build_rendered_site_check.py",
     "build_interactive_research_roadmap.py",
@@ -324,6 +326,7 @@ DOC_FILES = [
     "THREE_FOUNDATION_PIPELINES.md",
     "TWO_EVIDENCE_LINES.md",
     "TWO_EVIDENCE_LINE_RESULT_SUMMARY.md",
+    "QWEN3_OMNI_RUN_LINEAGE.md",
     "PROJECT_BRIEF.md",
     "PUBLIC_READER_MAP.md",
     "RENDERED_SITE_CHECK.md",