Upload 40 files

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easydense-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easydense-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac9543bf2faf5dbfb20c480bac3c3afc9be279ec156cfb5f03c1e70d9586305d
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easydense-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easymean-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easymean-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24fa28b232f7ae9a7a1a2408e71f847c3b039ec1c3d8932d94df67cd132d46db
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easymean-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easysparse-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easysparse-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9289c63384bc52b98c9640d91a00bd718910c43dff71e934456618cacf054ba
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-easysparse-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-harddense-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-harddense-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0c0d3efb072c6b2ef0809ce64d7543ed5d939b7394b43d6a741db0e83d4d1c7
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-harddense-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardmean-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardmean-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaa477e837e36c77fa7cff36292ef4915d9f8656342ae83fb25f9f0209bb87d9
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardmean-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardsparse-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardsparse-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef8263489f90229a59fbd39f82712f65c4e81183c1a78678ac0a969ebbda56b5
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-hardsparse-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumdense-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumdense-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b4f71a9b8cc7a4cd3c9b7f1ffabb3529ae17c8e1cfc7416a8ef2e7e9c0bbf42
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumdense-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easydense-v0",
+    "group": "OfflineMetadrive-easydense-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediummean-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediummean-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:655f5596257d25839439abe7c52bee11a7c5902a4f49dfe1dfca7ee3bbd3baef
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediummean-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easymean-v0",
+    "group": "OfflineMetadrive-easymean-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumsparse-v0_ddpm_lora_/ddpm_lora/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumsparse-v0_ddpm_lora_/ddpm_lora/model50000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c51a5ee4bf2317f64cd79e82f4bd2e7d718fa5eb4f36b1d1130607754bef610d
+size 41309313

LoRA-finetune/LoRA-rank8-alpha16-reward1-cost1/LoRA-rank8-alpha16-reward1-cost1_OfflineMetadrive-mediumsparse-v0_ddpm_lora_/ddpm_lora_bc/config.json

@@ -0,0 +1,69 @@
+{
+    "project": "PSEC",
+    "experiment_name": "ddpm_lora",
+    "timestamp": "LoRA-rank8-alpha16-reward1-cost1",
+    "max_steps": 1,
+    "pretrain_model": "",
+    "lora0": "",
+    "com_method": 0,
+    "batch_size": 2048,
+    "eval_episodes": 10,
+    "log_interval": 1000,
+    "save_steps": 50000,
+    "eval_interval": 50000,
+    "save_video": false,
+    "filter_threshold": null,
+    "take_top": null,
+    "online_max_steps": 0,
+    "unsquash_actions": false,
+    "normalize_returns": true,
+    "ratio": 1.0,
+    "training_time_inference_params": {
+        "N": 64,
+        "clip_sampler": true,
+        "M": 1
+    },
+    "rl_config": {
+        "model_cls": "LoRALearner",
+        "actor_lr": 0.0003,
+        "critic_lr": 0.0003,
+        "value_lr": 0.0003,
+        "T": 5,
+        "N": 64,
+        "M": 0,
+        "actor_dropout_rate": 0.1,
+        "actor_num_blocks": 3,
+        "decay_steps": 1000000,
+        "actor_layer_norm": true,
+        "value_layer_norm": true,
+        "actor_tau": 0.001,
+        "critic_objective": "expectile",
+        "critic_hyperparam": 0.7,
+        "cost_critic_hyperparam": 0.9,
+        "critic_type": "hj",
+        "cost_ub": 100,
+        "beta_schedule": "vp",
+        "cost_temperature": 1,
+        "reward_temperature": 1,
+        "cost_limit": 10,
+        "actor_objective": "bc",
+        "sampling_method": "ddpm",
+        "extract_method": "minqc",
+        "rank": 8,
+        "alpha_r": 16
+    },
+    "dataset_kwargs": {
+        "cost_scale": 25,
+        "pr_data": "data/point_robot-expert-random-100k.hdf5"
+    },
+    "seed": 42,
+    "env_name": "OfflineMetadrive-easysparse-v0",
+    "group": "OfflineMetadrive-easysparse-v0_ddpm_lora_",
+    "inference_variants": [
+        {
+            "N": 1,
+            "clip_sampler": true,
+            "M": 0
+        }
+    ]
+}