Index Medium and Strong trailing 4/8/16/32-update full models

Canonical update milestones map to zero-indexed iterations 3/7/15/31. All pointers reference public, exactly validated merged full-model repositories.

README.md

@@ -35,16 +35,21 @@ is available, still training, private, or not yet run.
 
 | PI | Teacher prompt | Iteration | Updates | Format | Direct load | Artifact |
 |---|---|---:|---:|---|---|---|
-| Medium | `trailing_user` | 8 | 9 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000008/artifact.json) |
-| Medium | `trailing_user` | 16 | 17 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000016/artifact.json) |
+| Medium | `trailing_user` | 3 | 4 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000003/artifact.json) |
+| Medium | `trailing_user` | 7 | 8 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000007/artifact.json) |
+| Medium | `trailing_user` | 15 | 16 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000015/artifact.json) |
 | Medium | `trailing_user` | 31 | 32 | merged full model | yes | [open](medium/trailing_user/iterations/iter_0000031/artifact.json) |
+| Strong | `trailing_user` | 3 | 4 | merged full model | yes | [open](strong/trailing_user/iterations/iter_0000003/artifact.json) |
+| Strong | `trailing_user` | 7 | 8 | merged full model | yes | [open](strong/trailing_user/iterations/iter_0000007/artifact.json) |
+| Strong | `trailing_user` | 15 | 16 | merged full model | yes | [open](strong/trailing_user/iterations/iter_0000015/artifact.json) |
+| Strong | `trailing_user` | 31 | 32 | merged full model | yes | [open](strong/trailing_user/iterations/iter_0000031/artifact.json) |
 | Weak | `trailing_user` | 1 | 2 | unmerged LoRA + full-MTP bundle | no | [open](weak/trailing_user/iterations/iter_0000001/artifact.json) |
 | Weak | `trailing_user` | 2 | 3 | unmerged LoRA + full-MTP bundle | no | [open](weak/trailing_user/iterations/iter_0000002/artifact.json) |
 | Weak | `trailing_user` | 3 | 4 | unmerged LoRA + full-MTP bundle | no | [open](weak/trailing_user/iterations/iter_0000003/artifact.json) |
 | Weak | `trailing_user` | 4 | 5 | unmerged LoRA + full-MTP bundle | no | [open](weak/trailing_user/iterations/iter_0000004/artifact.json) |
 
-Strong `trailing_user` is currently training. Its validated model repositories
-will be added without changing this hierarchy.
+The milestone labels count completed optimizer updates. The directory labels
+use zero-indexed iterations, so update 4/8/16/32 maps to iteration 3/7/15/31.
 
 Machine-readable inventory: [`registry.json`](registry.json). Artifact schema:
 [`schema/artifact.schema.json`](schema/artifact.schema.json).

medium/trailing_user/README.md

@@ -1,12 +1,13 @@
 # Medium PI / trailing_user
 
 Status: `available`. This arm contains the completed 1024-row, 32-update Medium
-PI training run and selected intermediate checkpoints.
+PI training run at the requested 4/8/16/32-update milestones.
 
 All indexed model repositories are fully merged and directly loadable. The PI
 was teacher-only during training; do not add Medium PI to the student prompt at
 evaluation time.
 
-- [`iter_0000008`](iterations/iter_0000008/artifact.json): 9 completed updates
-- [`iter_0000016`](iterations/iter_0000016/artifact.json): 17 completed updates
+- [`iter_0000003`](iterations/iter_0000003/artifact.json): 4 completed updates
+- [`iter_0000007`](iterations/iter_0000007/artifact.json): 8 completed updates
+- [`iter_0000015`](iterations/iter_0000015/artifact.json): 16 completed updates
 - [`iter_0000031`](iterations/iter_0000031/artifact.json): 32 completed updates, final

medium/trailing_user/iterations/{iter_0000008 → iter_0000003}/artifact.json

@@ -2,13 +2,13 @@
   "schema_version": 1,
   "pi_level": "medium",
   "teacher_prompt_mode": "trailing_user",
-  "source_iteration": 8,
+  "source_iteration": 3,
   "source_iteration_is_zero_indexed": true,
-  "completed_updates": 9,
+  "completed_updates": 4,
   "status": "available",
   "artifact_type": "merged_full_model",
-  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Iter8",
-  "revision": "cf3377168fc44582201c2453df73bcd50c3856f0",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Update04",
+  "revision": "71c3d258bcb5cd07f33d3afbc8a6266f8a2cafc3",
   "public": true,
   "directly_loadable": true,
   "includes": {

medium/trailing_user/iterations/{iter_0000016 → iter_0000007}/artifact.json

@@ -2,13 +2,13 @@
   "schema_version": 1,
   "pi_level": "medium",
   "teacher_prompt_mode": "trailing_user",
-  "source_iteration": 16,
+  "source_iteration": 7,
   "source_iteration_is_zero_indexed": true,
-  "completed_updates": 17,
+  "completed_updates": 8,
   "status": "available",
   "artifact_type": "merged_full_model",
-  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Iter16",
-  "revision": "4fdc187e0d6726d8870d2d5595deb0d4d21685fe",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Update08",
+  "revision": "0af9eba4227cf74f4b93e38e417d39625785eb31",
   "public": true,
   "directly_loadable": true,
   "includes": {

medium/trailing_user/iterations/iter_0000015/artifact.json

@@ -0,0 +1,22 @@
+{
+  "schema_version": 1,
+  "pi_level": "medium",
+  "teacher_prompt_mode": "trailing_user",
+  "source_iteration": 15,
+  "source_iteration_is_zero_indexed": true,
+  "completed_updates": 16,
+  "status": "available",
+  "artifact_type": "merged_full_model",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Update16",
+  "revision": "0237b456efe800fc2b7b4e589e6a62cd79d59f36",
+  "public": true,
+  "directly_loadable": true,
+  "includes": {
+    "merged_expert_sft_base": true,
+    "opsd_main_lora": true,
+    "opsd_mtp_lora": true,
+    "directly_trained_full_mtp": true,
+    "optimizer_state": false
+  },
+  "integrity_files": ["SHA256SUMS", "merge_manifest.json", "training_config.json"]
+}

medium/trailing_user/iterations/iter_0000031/artifact.json

@@ -7,11 +7,10 @@
   "completed_updates": 32,
   "status": "available",
   "artifact_type": "merged_full_model",
-  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged",
-  "revision": "545410ca91f4a3e8e7b93ccdb6deb782e4008a8b",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000-Merged-Update32",
+  "revision": "8a24e032fe35bce6e7854129a64bd8e5b772df7e",
   "public": true,
   "directly_loadable": true,
-  "supplemental_adapter_repo": "LSW142857/OPSD-PI-Qwen3.5-9B-Medium-Trailing-1024-A6000",
   "includes": {
     "merged_expert_sft_base": true,
     "opsd_main_lora": true,

registry.json

@@ -6,11 +6,14 @@
   "teacher_prompt_modes": ["first_user", "trailing_user"],
   "arms": {
     "strong/first_user": {"status": "not_indexed", "iterations": []},
-    "strong/trailing_user": {"status": "training", "iterations": []},
+    "strong/trailing_user": {
+      "status": "available",
+      "iterations": [3, 7, 15, 31]
+    },
     "medium/first_user": {"status": "not_indexed", "iterations": []},
     "medium/trailing_user": {
       "status": "available",
-      "iterations": [8, 16, 31]
+      "iterations": [3, 7, 15, 31]
     },
     "weak/first_user": {"status": "private_not_indexed", "iterations": []},
     "weak/trailing_user": {
@@ -19,9 +22,14 @@
     }
   },
   "artifacts": [
-    "medium/trailing_user/iterations/iter_0000008/artifact.json",
-    "medium/trailing_user/iterations/iter_0000016/artifact.json",
+    "medium/trailing_user/iterations/iter_0000003/artifact.json",
+    "medium/trailing_user/iterations/iter_0000007/artifact.json",
+    "medium/trailing_user/iterations/iter_0000015/artifact.json",
     "medium/trailing_user/iterations/iter_0000031/artifact.json",
+    "strong/trailing_user/iterations/iter_0000003/artifact.json",
+    "strong/trailing_user/iterations/iter_0000007/artifact.json",
+    "strong/trailing_user/iterations/iter_0000015/artifact.json",
+    "strong/trailing_user/iterations/iter_0000031/artifact.json",
     "weak/trailing_user/iterations/iter_0000001/artifact.json",
     "weak/trailing_user/iterations/iter_0000002/artifact.json",
     "weak/trailing_user/iterations/iter_0000003/artifact.json",

strong/trailing_user/README.md

@@ -1,7 +1,7 @@
 # Strong PI / trailing_user
 
-Status: `training`. The 1024-row, 32-update Strong PI `trailing_user` production
-run is still in progress. Validated checkpoints will be registered below after
-their model repositories are uploaded and verified.
+Status: `available`. The 1024-row, 32-update Strong PI `trailing_user`
+production run is complete. The requested 4/8/16/32-update milestones are
+fully merged, public, and directly loadable.
 
 Iteration directory: [`iterations`](iterations/README.md).

strong/trailing_user/iterations/README.md

@@ -1,4 +1,8 @@
 # Iterations
 
-Training is in progress. This directory intentionally contains no pointer to a
-local-only checkpoint; only validated Hugging Face artifacts are indexed.
+Only validated, public Hugging Face full-model artifacts are indexed:
+
+- [`iter_0000003`](iter_0000003/artifact.json): 4 completed updates
+- [`iter_0000007`](iter_0000007/artifact.json): 8 completed updates
+- [`iter_0000015`](iter_0000015/artifact.json): 16 completed updates
+- [`iter_0000031`](iter_0000031/artifact.json): 32 completed updates, final

strong/trailing_user/iterations/iter_0000003/artifact.json

@@ -0,0 +1,22 @@
+{
+  "schema_version": 1,
+  "pi_level": "strong",
+  "teacher_prompt_mode": "trailing_user",
+  "source_iteration": 3,
+  "source_iteration_is_zero_indexed": true,
+  "completed_updates": 4,
+  "status": "available",
+  "artifact_type": "merged_full_model",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Strong-Trailing-1024-A6000-Merged-Update04",
+  "revision": "f2201db448197cca8b384bfa4c4ba9209bb63308",
+  "public": true,
+  "directly_loadable": true,
+  "includes": {
+    "merged_expert_sft_base": true,
+    "opsd_main_lora": true,
+    "opsd_mtp_lora": true,
+    "directly_trained_full_mtp": true,
+    "optimizer_state": false
+  },
+  "integrity_files": ["SHA256SUMS", "merge_manifest.json", "training_config.json"]
+}

strong/trailing_user/iterations/iter_0000007/artifact.json

@@ -0,0 +1,22 @@
+{
+  "schema_version": 1,
+  "pi_level": "strong",
+  "teacher_prompt_mode": "trailing_user",
+  "source_iteration": 7,
+  "source_iteration_is_zero_indexed": true,
+  "completed_updates": 8,
+  "status": "available",
+  "artifact_type": "merged_full_model",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Strong-Trailing-1024-A6000-Merged-Update08",
+  "revision": "f122ec82e82789ab2e67424cc6f01cb8ef452e0a",
+  "public": true,
+  "directly_loadable": true,
+  "includes": {
+    "merged_expert_sft_base": true,
+    "opsd_main_lora": true,
+    "opsd_mtp_lora": true,
+    "directly_trained_full_mtp": true,
+    "optimizer_state": false
+  },
+  "integrity_files": ["SHA256SUMS", "merge_manifest.json", "training_config.json"]
+}

strong/trailing_user/iterations/iter_0000015/artifact.json

@@ -0,0 +1,22 @@
+{
+  "schema_version": 1,
+  "pi_level": "strong",
+  "teacher_prompt_mode": "trailing_user",
+  "source_iteration": 15,
+  "source_iteration_is_zero_indexed": true,
+  "completed_updates": 16,
+  "status": "available",
+  "artifact_type": "merged_full_model",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Strong-Trailing-1024-A6000-Merged-Update16",
+  "revision": "e208ee6caf07b4a9e4c03b3c34368380a8c782f3",
+  "public": true,
+  "directly_loadable": true,
+  "includes": {
+    "merged_expert_sft_base": true,
+    "opsd_main_lora": true,
+    "opsd_mtp_lora": true,
+    "directly_trained_full_mtp": true,
+    "optimizer_state": false
+  },
+  "integrity_files": ["SHA256SUMS", "merge_manifest.json", "training_config.json"]
+}

strong/trailing_user/iterations/iter_0000031/artifact.json

@@ -0,0 +1,22 @@
+{
+  "schema_version": 1,
+  "pi_level": "strong",
+  "teacher_prompt_mode": "trailing_user",
+  "source_iteration": 31,
+  "source_iteration_is_zero_indexed": true,
+  "completed_updates": 32,
+  "status": "available",
+  "artifact_type": "merged_full_model",
+  "repo_id": "LSW142857/OPSD-PI-Qwen3.5-9B-Strong-Trailing-1024-A6000-Merged-Update32",
+  "revision": "44a95f1e74cbba504b2d0a8edc2fc1e6ae0d5bca",
+  "public": true,
+  "directly_loadable": true,
+  "includes": {
+    "merged_expert_sft_base": true,
+    "opsd_main_lora": true,
+    "opsd_mtp_lora": true,
+    "directly_trained_full_mtp": true,
+    "optimizer_state": false
+  },
+  "integrity_files": ["SHA256SUMS", "merge_manifest.json", "training_config.json"]
+}