Added the base checkpoint

features.json

@@ -0,0 +1,38 @@
+{
+  "action": {
+    "type": "ACTION",
+    "shape": [
+      20
+    ]
+  },
+  "observation.image.0": {
+    "type": "VISUAL",
+    "shape": [
+      3,
+      224,
+      224
+    ]
+  },
+  "observation.image.1": {
+    "type": "VISUAL",
+    "shape": [
+      3,
+      224,
+      224
+    ]
+  },
+  "observation.image.2": {
+    "type": "VISUAL",
+    "shape": [
+      3,
+      224,
+      224
+    ]
+  },
+  "observation.state": {
+    "type": "STATE",
+    "shape": [
+      20
+    ]
+  }
+}

manifest.json

@@ -0,0 +1,22 @@
+{
+  "format": "rho",
+  "version": 1,
+  "step": 90000,
+  "weights": {
+    "format": "safetensors",
+    "max_shard_size": "5GB"
+  },
+  "artifacts": {
+    "policy": "policy.json",
+    "features": "features.json",
+    "data_config": "data_config.json",
+    "stats": null,
+    "training_state": null
+  },
+  "save_timings_seconds": {
+    "model_weights": 3.0034489850513637,
+    "metadata": 0.015505407936871052,
+    "training_state": null,
+    "total_before_publish": 3.0189577320124954
+  }
+}

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fda061a1fc5b2535a78ffc7a55a502f9fd1c0cf4e663826202b1fdac875f477
+size 4995666736

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:810f7e18b3bc8ed74f406d07f96f84e94f57c874fcdde2d098e6145d87f1068c
+size 4985052104

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28ec20ef05a4d5dcaa237fd45778405915ba699ac15e96bd630bbdc64a855742
+size 520742056

policy.json

@@ -0,0 +1,75 @@
+{
+  "type": "rho1",
+  "name": "rho1",
+  "device": "cuda",
+  "dtype": "bfloat16",
+  "lr_scheduler": {
+    "type": "warmup_stable_decay",
+    "num_warmup_steps": 1000,
+    "num_decay_steps": 45000,
+    "peak_lr": 0.0001,
+    "decay_lr": 1e-05,
+    "num_cycles": 1,
+    "num_rewarm_steps": null,
+    "name": "warmup_stable_decay"
+  },
+  "optimizer": {
+    "type": "adamw",
+    "lr": 1e-06,
+    "weight_decay": 0.1,
+    "grad_clip_norm": 10.0,
+    "betas": [
+      0.9,
+      0.95
+    ],
+    "eps": 1e-06
+  },
+  "resize_imgs_with_padding": [
+    224,
+    224
+  ],
+  "embed_dim": 2048,
+  "num_heads": 16,
+  "ff_dim": 4096,
+  "norm": "adaptive",
+  "hidden_state_idx": 14,
+  "num_blocks": 12,
+  "max_seq_len": 6144,
+  "n_obs_steps": 1,
+  "chunk_size": 50,
+  "n_action_steps": 25,
+  "max_state_dim": 32,
+  "max_action_dim": 32,
+  "empty_cameras": 0,
+  "num_steps": 10,
+  "num_flow_samples": 1,
+  "attention_implementation": "flash_attention_2",
+  "attention_type": "cross",
+  "adaln_mode": "shared",
+  "adaln_lora_rank": 256,
+  "gqa_groups": 4,
+  "cross_attention_dim": null,
+  "vlm_layer_select": "last",
+  "shared_vlm_projection": false,
+  "kv_pos_encoding": true,
+  "freeze_vlm_backbone": false,
+  "freeze_vision_encoder": false,
+  "freeze_language_model": false,
+  "train_expert_only": false,
+  "enable_gradient_checkpointing": true,
+  "optimizer_lr": 1e-06,
+  "optimizer_lr_action_expert": 0.0001,
+  "optimizer_betas": [
+    0.9,
+    0.95
+  ],
+  "optimizer_eps": 1e-06,
+  "optimizer_weight_decay": 0.1,
+  "scheduler_warmup_steps": 1000,
+  "scheduler_decay_steps": 30000,
+  "scheduler_decay_lr": 2.5e-06,
+  "time_sampling_strategy": "beta",
+  "dropout_p": 0.1,
+  "pos_emb_method": "sinusoidal",
+  "log_hidden_state_stats": true
+}