test-1-checkpoint-1500

test-1-checkpoint-1500/config.json

@@ -0,0 +1,163 @@
+{
+  "action_horizon": 50,
+  "batch_size": 8,
+  "buffer_size": 64,
+  "eval_mix": "qiwu_test",
+  "frozen_keys": [
+    "*hf_model*"
+  ],
+  "lora": false,
+  "lora_config": {
+    "bias": "none",
+    "lora_alpha": 16,
+    "lora_dropout": 0.05,
+    "r": 8
+  },
+  "model": {
+    "heads": {
+      "action": {
+        "args": [],
+        "kwargs": {
+          "action_dim": 7,
+          "action_horizon": 50,
+          "readout_key": "readout_action",
+          "token_embedding_size": 768
+        },
+        "module": "model.octo.components.action_heads",
+        "name": "L1ActionHead"
+      }
+    },
+    "max_horizon": 10,
+    "observation_tokenizers": {
+      "primary": {
+        "args": [],
+        "kwargs": {
+          "encoder": {
+            "args": [],
+            "kwargs": {
+              "in_features": 6
+            },
+            "module": "octo.model.components.vit_encoders",
+            "name": "SmallStem16"
+          },
+          "obs_stack_keys": [
+            "image_primary"
+          ],
+          "task_stack_keys": [
+            "image_primary"
+          ]
+        },
+        "module": "octo.model.components.tokenizers",
+        "name": "ImageTokenizer"
+      },
+      "state": {
+        "args": [],
+        "kwargs": {
+          "bin_type": "normal",
+          "high": 2.0,
+          "low": -2.0,
+          "n_bins": 256,
+          "obs_keys": [
+            "state"
+          ]
+        },
+        "module": "model.octo.components.tokenizers",
+        "name": "LowdimObsTokenizer"
+      }
+    },
+    "readouts": {
+      "action": 1
+    },
+    "repeat_task_tokens": true,
+    "task_tokenizers": {
+      "language": {
+        "args": [],
+        "kwargs": {
+          "encoder": "t5-base",
+          "finetune_encoder": false
+        },
+        "module": "octo.model.components.tokenizers",
+        "name": "LanguageTokenizer"
+      }
+    },
+    "token_embedding_size": 768,
+    "transformer_kwargs": {
+      "add_position_embedding": false,
+      "attention_dropout_rate": 0.0,
+      "dropout_rate": 0.0,
+      "mlp_dim": 3072,
+      "num_attention_heads": 12,
+      "num_layers": 12
+    },
+    "use_correct_attention": true
+  },
+  "no_state": false,
+  "obs_token_nums": {
+    "primary": 256,
+    "state": 7
+  },
+  "output_dir": "qiwu_checkpoints4",
+  "pretrained": {
+    "kwargs": {
+      "subpath": "oxe-g2-checkpoint-300000"
+    },
+    "model": "hf://chuanmew/octo_torch"
+  },
+  "run_name": "qiwu",
+  "sampler_num_samples": 2560000,
+  "seed": 42,
+  "subsample_length": 99999,
+  "text_processor": {
+    "args": [],
+    "kwargs": {
+      "encode_with_model": false,
+      "tokenizer_kwargs": {
+        "max_length": 16,
+        "padding": "max_length",
+        "return_tensors": "np",
+        "truncation": true
+      },
+      "tokenizer_name": "t5-base"
+    },
+    "module": "octo.components.text_processing",
+    "name": "HFTokenizer"
+  },
+  "train_mix": "qiwu_train",
+  "training_arguments": {
+    "bf16": true,
+    "dataloader_pin_memory": true,
+    "ddp_find_unused_parameters": true,
+    "eval_steps": 100,
+    "gradient_accumulation_steps": 1,
+    "learning_rate": 3e-05,
+    "logging_nan_inf_filter": false,
+    "logging_steps": 100,
+    "lr_scheduler_kwargs": {
+      "decay_type": "cosine",
+      "min_lr_ratio": 0.05,
+      "num_decay_steps": 0
+    },
+    "lr_scheduler_type": "warmup_stable_decay",
+    "max_grad_norm": 1.0,
+    "max_steps": 300000,
+    "optim": "adamw_torch_fused",
+    "per_device_eval_batch_size": 1,
+    "per_device_train_batch_size": 1,
+    "report_to": "wandb",
+    "save_safetensors": true,
+    "save_steps": 500,
+    "torch_compile": true,
+    "warmup_steps": 100,
+    "weight_decay": 0.1
+  },
+  "training_keys": [
+    "*action*",
+    "*observation_tokenizers*",
+    "*transformer.transformer*",
+    "*task_projections*",
+    "*obs_projections*",
+    "*readout_embeddings*",
+    "*task_pos_embeddings*",
+    "*obs_pos_embeddings*"
+  ]
+}

test-1-checkpoint-1500/dataset_statistics.json

@@ -0,0 +1,98 @@
+[
+    {
+        "action": {
+            "mean": [
+                0.005713334307074547,
+                -8.683370651851874e-06,
+                -0.0027405666187405586,
+                -0.01428219024091959,
+                -0.00569964898750186,
+                0.03178633004426956,
+                0.31847861409187317
+            ],
+            "std": [
+                0.017133653163909912,
+                0.012163525447249413,
+                0.028012460097670555,
+                0.056618135422468185,
+                0.059768617153167725,
+                0.08718060702085495,
+                0.45297929644584656
+            ],
+            "min": [
+                -0.045306507498025894,
+                -0.058269500732421875,
+                -0.0945921391248703,
+                -0.39226841926574707,
+                -0.31778502464294434,
+                -0.46861958503723145,
+                0.0
+            ],
+            "max": [
+                0.10087767243385315,
+                0.06844352185726166,
+                0.13764026761054993,
+                0.39099645614624023,
+                0.4911177158355713,
+                0.7427346706390381,
+                1.0
+            ],
+            "mask": [
+                true,
+                true,
+                true,
+                true,
+                true,
+                true,
+                false
+            ]
+        },
+        "state": {
+            "mean": [
+                -1.2245632410049438,
+                -1.050716519355774,
+                -1.258064866065979,
+                -0.5181024670600891,
+                -0.9791191816329956,
+                0.11057784408330917,
+                0.26232674717903137
+            ],
+            "std": [
+                0.23113861680030823,
+                0.32922741770744324,
+                0.2693513333797455,
+                0.32120275497436523,
+                0.2501857280731201,
+                0.4673791527748108,
+                0.42628398537635803
+            ],
+            "min": [
+                -1.8634008169174194,
+                -1.9834270477294922,
+                -1.991769790649414,
+                -1.7809165716171265,
+                -1.5987040996551514,
+                -1.2469679117202759,
+                0.0
+            ],
+            "max": [
+                -0.6335545182228088,
+                -0.20619319379329681,
+                -0.40194931626319885,
+                0.5385388135910034,
+                0.03621558099985123,
+                1.5524179935455322,
+                1.0
+            ],
+            "mask": [
+                true,
+                true,
+                true,
+                true,
+                true,
+                true,
+                false
+            ]
+        }
+    }
+]

test-1-checkpoint-1500/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:226a8bda34b980c3c87f83da6748f0ae0fa8cb5bcd2627d30b9179bbfd8a32a7
+size 412295788

test-1-checkpoint-1500/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e128857b241bc01a1db11900bc9169b249cff6c24549645d4251a94614e59226
+size 386143930

test-1-checkpoint-1500/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:916ccac3d8b32b81123571dde2d9dbe2f1f07ebd90ef54a304aaf7c937ef1a95
+size 15984

test-1-checkpoint-1500/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:333aef0819b926923bec9f240bca2466cd68a800ab2e9ab4c6e2335f27143f83
+size 15984

test-1-checkpoint-1500/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81f770f6f8cab2724d297e3f5e935dc10562e83c9b4a2bafe5cacde08d67d220
+size 15984

test-1-checkpoint-1500/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b2a15f877a0d8e205b82d2ee3a0daaa5b69a41a8787ae96d50447e7fd8616a4
+size 15984

test-1-checkpoint-1500/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50dce3f47a9c4d3baf720ac97339bde0244e272c039c1780d2fce53bdc309080
+size 15984

test-1-checkpoint-1500/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d92fd2e6d58ec26cc909989125659fa97e9aa17e622cefad055005be36d6b81a
+size 15984

test-1-checkpoint-1500/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c306d4a4f506e6752ae90755d2e4fcb4604eea922888c2e19758160d22b62474
+size 15984

test-1-checkpoint-1500/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab9712425612fc7ead06367303ef37eacf7dcc969665997a0cac0f28a009a04a
+size 15984

test-1-checkpoint-1500/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed4336dbf0ba1587dcd2e69dc77b38f8ff64571073dd364cce7a4f08ec4c85e5
+size 1064

test-1-checkpoint-1500/trainer_state.json

@@ -0,0 +1,259 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0375,
+  "eval_steps": 100,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0025,
+      "grad_norm": 4.0,
+      "learning_rate": 2.9715000000000003e-05,
+      "loss": 4.4555,
+      "step": 100
+    },
+    {
+      "epoch": 0.0025,
+      "eval_loss": 4.083842754364014,
+      "eval_runtime": 15.1972,
+      "eval_samples_per_second": 1330.307,
+      "eval_steps_per_second": 166.346,
+      "step": 100
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 6.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7452,
+      "step": 200
+    },
+    {
+      "epoch": 0.005,
+      "eval_loss": 3.76338529586792,
+      "eval_runtime": 14.6937,
+      "eval_samples_per_second": 1375.896,
+      "eval_steps_per_second": 172.047,
+      "step": 200
+    },
+    {
+      "epoch": 0.0075,
+      "grad_norm": 10.875,
+      "learning_rate": 3e-05,
+      "loss": 3.5303,
+      "step": 300
+    },
+    {
+      "epoch": 0.0075,
+      "eval_loss": 3.6259424686431885,
+      "eval_runtime": 14.8394,
+      "eval_samples_per_second": 1362.388,
+      "eval_steps_per_second": 170.357,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.3359,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 3.4924354553222656,
+      "eval_runtime": 14.869,
+      "eval_samples_per_second": 1359.679,
+      "eval_steps_per_second": 170.019,
+      "step": 400
+    },
+    {
+      "epoch": 0.0125,
+      "grad_norm": 7.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.2483,
+      "step": 500
+    },
+    {
+      "epoch": 0.0125,
+      "eval_loss": 3.427826404571533,
+      "eval_runtime": 14.83,
+      "eval_samples_per_second": 1363.252,
+      "eval_steps_per_second": 170.466,
+      "step": 500
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 6.75,
+      "learning_rate": 3e-05,
+      "loss": 3.174,
+      "step": 600
+    },
+    {
+      "epoch": 0.015,
+      "eval_loss": 3.4111289978027344,
+      "eval_runtime": 14.9413,
+      "eval_samples_per_second": 1353.099,
+      "eval_steps_per_second": 169.196,
+      "step": 600
+    },
+    {
+      "epoch": 0.0175,
+      "grad_norm": 9.125,
+      "learning_rate": 3e-05,
+      "loss": 3.0925,
+      "step": 700
+    },
+    {
+      "epoch": 0.0175,
+      "eval_loss": 3.364833354949951,
+      "eval_runtime": 14.9006,
+      "eval_samples_per_second": 1356.79,
+      "eval_steps_per_second": 169.657,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.0564,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 3.3411409854888916,
+      "eval_runtime": 14.9667,
+      "eval_samples_per_second": 1350.796,
+      "eval_steps_per_second": 168.908,
+      "step": 800
+    },
+    {
+      "epoch": 0.0225,
+      "grad_norm": 6.34375,
+      "learning_rate": 3e-05,
+      "loss": 2.9535,
+      "step": 900
+    },
+    {
+      "epoch": 0.0225,
+      "eval_loss": 3.31355881690979,
+      "eval_runtime": 14.9497,
+      "eval_samples_per_second": 1352.334,
+      "eval_steps_per_second": 169.1,
+      "step": 900
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 6.375,
+      "learning_rate": 3e-05,
+      "loss": 2.9478,
+      "step": 1000
+    },
+    {
+      "epoch": 0.025,
+      "eval_loss": 3.3087844848632812,
+      "eval_runtime": 14.8619,
+      "eval_samples_per_second": 1360.323,
+      "eval_steps_per_second": 170.099,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0275,
+      "grad_norm": 13.0,
+      "learning_rate": 3e-05,
+      "loss": 2.876,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0275,
+      "eval_loss": 3.295400381088257,
+      "eval_runtime": 15.1916,
+      "eval_samples_per_second": 1330.805,
+      "eval_steps_per_second": 166.408,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.75,
+      "learning_rate": 3e-05,
+      "loss": 2.8775,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.292489767074585,
+      "eval_runtime": 14.7952,
+      "eval_samples_per_second": 1366.461,
+      "eval_steps_per_second": 170.867,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0325,
+      "grad_norm": 14.25,
+      "learning_rate": 3e-05,
+      "loss": 2.8152,
+      "step": 1300
+    },
+    {
+      "epoch": 0.0325,
+      "eval_loss": 3.2735142707824707,
+      "eval_runtime": 14.9427,
+      "eval_samples_per_second": 1352.969,
+      "eval_steps_per_second": 169.18,
+      "step": 1300
+    },
+    {
+      "epoch": 0.035,
+      "grad_norm": 8.25,
+      "learning_rate": 3e-05,
+      "loss": 2.7434,
+      "step": 1400
+    },
+    {
+      "epoch": 0.035,
+      "eval_loss": 3.2721683979034424,
+      "eval_runtime": 14.9214,
+      "eval_samples_per_second": 1354.903,
+      "eval_steps_per_second": 169.422,
+      "step": 1400
+    },
+    {
+      "epoch": 0.0375,
+      "grad_norm": 8.625,
+      "learning_rate": 3e-05,
+      "loss": 2.7628,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0375,
+      "eval_loss": 3.264232873916626,
+      "eval_runtime": 14.791,
+      "eval_samples_per_second": 1366.845,
+      "eval_steps_per_second": 170.915,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 300000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

test-1-checkpoint-1500/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dbd37b641e7bea235d5fc9c06dc239b106449ff9247ac5d78f46fac43e0b9ad
+size 5432