Release initial Double-DQN consolidation checkpoint

Add the initial 71-128-64-1 consolidation policy, configuration, model card, and SHA256 checksums.

README.md

@@ -0,0 +1,70 @@
+---
+tags:
+  - robotics
+  - humanoid
+  - reinforcement-learning
+  - double-dqn
+  - pytorch
+---
+
+# HumanoidTTT — Consolidation Policy
+
+Initial Double-DQN consolidation weights for
+[HumaniodTTT: Test-Time Capability Reuse for Efficient Humanoid Control](https://github.com/AIGeeksGroup/HumaniodTTT).
+
+## Model
+
+The policy scores retention actions for a finite-capacity capability store. Each
+action has a 71-dimensional consolidation feature and is scored by a shared
+`71 → 128 → 64 → 1` ReLU network with **17,537 parameters**. When the ten-slot store
+is full, the actions are `SKIP` or `REPLACE(j)`. Online Double-DQN updates use the
+fraction of subsequent requests with successful reuse between consecutive
+full-store qualified-miss decisions.
+
+## Release files
+
+| File | Contents |
+| --- | --- |
+| `consolidation_policy.pt` | Initial FP32 PyTorch scorer state dictionary, seed 83001. |
+| `config.json` | Architecture, feature layout, action mapping, online settings, and weight checksum. |
+| `SHA256SUMS` | File checksums. |
+
+This checkpoint is the starting point before online adaptation. It does not
+contain a prefilled capability store or optimizer/replay state.
+
+## Usage
+
+Install the code and download dependencies:
+
+```bash
+git clone https://github.com/AIGeeksGroup/HumaniodTTT.git
+cd HumaniodTTT
+pip install -e .
+pip install huggingface_hub
+```
+
+```python
+import torch
+from huggingface_hub import hf_hub_download
+from humanoid_ttt import ConsolidationPolicy
+
+path = hf_hub_download("AIGeeksGroup/HumanoidTTT", "consolidation_policy.pt")
+weights = torch.load(path, map_location="cpu", weights_only=True)
+policy = ConsolidationPolicy(weights, capacity=10, online_learning_rate=3e-5, gamma=0.95)
+```
+
+See the [GitHub README](https://github.com/AIGeeksGroup/HumaniodTTT#using-the-components)
+for entry applicability, execution feedback, and update interfaces. Pin the model
+revision and code commit when reproducing experiments.
+
+## Dependencies and scope
+
+The 45-dimensional A2 entry-applicability module uses state features and geometric
+certificates; it has no separate neural-network checkpoint. Applications supply
+motion-specific certificates, qualification evidence, and the execution runtime.
+Download frozen generation and tracking models from
+[OMG](https://github.com/Tsinghua-MARS-Lab/OMG) and
+[HoloMotion](https://github.com/HorizonRobotics/HoloMotion).
+
+The released scorer weights alone do not constitute an end-to-end robot controller.
+Successful loading verifies network compatibility, not complete experimental reproduction.

SHA256SUMS

@@ -0,0 +1,3 @@
+2b9232271b1c1e64393a687104570bd04508f39c6abc96baaae4c0fd528072fc  config.json
+cdb1b5f4f9c671da5c6c26f9d9cc752aee3c81fc87b40dad67d0efa02f0e60b5  consolidation_policy.pt
+8093b0bb8d7cfbd284b7ab89fcb89453f8842a0ed9332910d30da39da819ef93  README.md

config.json

@@ -0,0 +1,97 @@
+{
+  "format_version": 1,
+  "algorithm": "Double-DQN",
+  "checkpoint": "consolidation_policy.pt",
+  "checkpoint_sha256": "cdb1b5f4f9c671da5c6c26f9d9cc752aee3c81fc87b40dad67d0efa02f0e60b5",
+  "checkpoint_stage": "before_online_adaptation",
+  "initialization_seed": 83001,
+  "architecture": [
+    71,
+    128,
+    64,
+    1
+  ],
+  "hidden_activation": "ReLU",
+  "parameter_count": 17537,
+  "dtype": "float32",
+  "state_dict_format": "raw_pytorch_state_dict",
+  "capacity": 10,
+  "recent_request_window": 16,
+  "action_count_when_full": 11,
+  "actions": {
+    "0": "SKIP",
+    "1..10": "REPLACE(entries[action_index - 1])"
+  },
+  "api_action_labels": {
+    "SKIP": "DROP",
+    "REPLACE": "REPLACE",
+    "free_slot": "INSERT"
+  },
+  "feature_layout": [
+    {
+      "slice": [
+        0,
+        16
+      ],
+      "name": "candidate_motion_descriptor"
+    },
+    {
+      "slice": [
+        16,
+        32
+      ],
+      "name": "store_descriptor_mean"
+    },
+    {
+      "slice": [
+        32,
+        48
+      ],
+      "name": "store_descriptor_max"
+    },
+    {
+      "slice": [
+        48,
+        64
+      ],
+      "name": "replacement_target_descriptor_or_zeros_for_skip"
+    },
+    {
+      "slice": [
+        64,
+        69
+      ],
+      "name": "global_features",
+      "fields": [
+        "store_size_div_20",
+        "recent_unique_count_div_16",
+        "candidate_recent_frequency",
+        "mean_candidate_store_distance_div_20",
+        "min_full_store_decisions_20_div_20"
+      ]
+    },
+    {
+      "slice": [
+        69,
+        71
+      ],
+      "name": "action_features",
+      "fields": [
+        "is_skip",
+        "target_recent_frequency"
+      ]
+    }
+  ],
+  "online_learning_rate": 3e-05,
+  "gamma": 0.95,
+  "optimizer": "Adam",
+  "loss": "smooth_l1",
+  "gradient_norm_clip": 5.0,
+  "target_sync_optimizer_steps": 50,
+  "replay_min_transitions": 8,
+  "replay_draws_per_update_call": 8,
+  "reward": "successful reuse count / request count between consecutive full-store qualified-miss decisions",
+  "reward_input": "observe_request(capability_id, reused=successful_reuse)",
+  "code_repository": "https://github.com/AIGeeksGroup/HumaniodTTT",
+  "verified_source_commit": "39f33575fff8378c0cabf4571b595646d833667c"
+}

consolidation_policy.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdb1b5f4f9c671da5c6c26f9d9cc752aee3c81fc87b40dad67d0efa02f0e60b5
+size 73154