Upload lewm-rs training artifacts

train/pusht-short-20260514T114211Z/step_0000010.json

@@ -0,0 +1,30 @@
+{
+  "schema_version": "1.0",
+  "run_id": "pusht-action-probe-v1",
+  "step": 10,
+  "epoch": 0,
+  "wall_time_s": 0.0,
+  "git_short_sha": "unknown",
+  "config_hash": "a7089ee31af7",
+  "rng_state": {
+    "global_seed": 0,
+    "step_at_save": 10,
+    "data_shuffle": "sequential-window-modulo-v1",
+    "sigreg_sketch": "disabled-for-action-probe",
+    "dropout": "disabled-for-action-probe",
+    "cem": "disabled-for-action-probe",
+    "model_init": "pusht-action-probe-zero-init"
+  },
+  "metrics_last_step": {
+    "loss/train": 0.037582458725922085,
+    "optim/grad_norm_post": 0.18590227536927145,
+    "optim/grad_norm_pre": 0.18590227536927145,
+    "train/grad_explosion_events": 0.0,
+    "train/samples_seen": 640.0
+  },
+  "checkpoint_files": {
+    "model_burn": "step_0000010.mpk",
+    "model_safetensors": "step_0000010.safetensors",
+    "parity": "step_0000010.parity.json"
+  }
+}

train/pusht-short-20260514T114211Z/step_0000010.mpk

@@ -0,0 +1 @@
+{"schema_version":"1.0.0","kind":"lewm-rs-pusht-action-probe-record","step":10,"params":[-0.000013179592337080288,0.000013473546857790225,-0.000013181912014974756,0.00001347062919911206,4.031473247576911e-6,5.9841253875691976e-6],"adamw_step":10,"samples_seen":640}

train/pusht-short-20260514T114211Z/step_0000010.parity.json

@@ -0,0 +1,5 @@
+{
+  "encoder_cls_l_inf": 0.0,
+  "predictor_l_inf": 0.0,
+  "sigreg_value": 0.037582458725922085
+}

train/pusht-short-20260514T114211Z/step_0000010.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d87ab1b2c6be579adb8c8ecb59a8358f45edd919fec25270bc2efd9a1e70d1e7
+size 256

train/pusht-short-20260514T114211Z/train_losses.jsonl

@@ -0,0 +1,10 @@
+{"step":1,"loss":0.03653029702945655,"grad_norm_pre":0.21102662425226687,"grad_norm_post":0.21102662425226687,"learning_rate":3e-7,"samples_seen":64}
+{"step":2,"loss":0.03820481101365848,"grad_norm_pre":0.1665218991176013,"grad_norm_post":0.1665218991176013,"learning_rate":6e-7,"samples_seen":128}
+{"step":3,"loss":0.02324381511204751,"grad_norm_pre":0.06870479521984989,"grad_norm_post":0.06870479521984989,"learning_rate":9e-7,"samples_seen":192}
+{"step":4,"loss":0.03268800624102926,"grad_norm_pre":0.15994647974291268,"grad_norm_post":0.15994647974291268,"learning_rate":1.2e-6,"samples_seen":256}
+{"step":5,"loss":0.03540466956607878,"grad_norm_pre":0.07796946867216864,"grad_norm_post":0.07796946867216864,"learning_rate":1.4999999999999998e-6,"samples_seen":320}
+{"step":6,"loss":0.027722052435277388,"grad_norm_pre":0.11633184139539246,"grad_norm_post":0.11633184139539246,"learning_rate":1.8e-6,"samples_seen":384}
+{"step":7,"loss":0.03432064541904802,"grad_norm_pre":0.17371458729824793,"grad_norm_post":0.17371458729824793,"learning_rate":2.1e-6,"samples_seen":448}
+{"step":8,"loss":0.0243554438932299,"grad_norm_pre":0.05178564291500989,"grad_norm_post":0.05178564291500989,"learning_rate":2.4e-6,"samples_seen":512}
+{"step":9,"loss":0.03652638504573583,"grad_norm_pre":0.1883752418556006,"grad_norm_post":0.1883752418556006,"learning_rate":2.6999999999999996e-6,"samples_seen":576}
+{"step":10,"loss":0.037582458725922085,"grad_norm_pre":0.18590227536927145,"grad_norm_post":0.18590227536927145,"learning_rate":2.9999999999999997e-6,"samples_seen":640}

train/pusht-short-20260514T114211Z/train_report.json

@@ -0,0 +1,109 @@
+{
+  "schema_version": "1.0.0",
+  "kind": "lewm-rs-train-report",
+  "config_hash": "a7089ee31af7",
+  "output_dir": "/tmp/out",
+  "data_dir": "/tmp/data",
+  "data_source": "pusht-hdf5:/tmp/data",
+  "dataset_windows": 2092476,
+  "max_steps": 10,
+  "steps_completed": 10,
+  "batch_size": 64,
+  "seed": 0,
+  "device": "cuda:0",
+  "initial_loss": 0.03653029702945655,
+  "final_loss": 0.037582458725922085,
+  "loss_decreased": false,
+  "checkpoint_step": 10,
+  "checkpoint_complete": true,
+  "checkpoint_files": [
+    "step_0000010.mpk",
+    "step_0000010.safetensors",
+    "step_0000010.json",
+    "step_0000010.parity.json"
+  ],
+  "grad_explosion_events": 0,
+  "mode": "pusht-action-probe",
+  "losses": [
+    {
+      "step": 1,
+      "loss": 0.03653029702945655,
+      "grad_norm_pre": 0.21102662425226687,
+      "grad_norm_post": 0.21102662425226687,
+      "learning_rate": 3e-7,
+      "samples_seen": 64
+    },
+    {
+      "step": 2,
+      "loss": 0.03820481101365848,
+      "grad_norm_pre": 0.1665218991176013,
+      "grad_norm_post": 0.1665218991176013,
+      "learning_rate": 6e-7,
+      "samples_seen": 128
+    },
+    {
+      "step": 3,
+      "loss": 0.02324381511204751,
+      "grad_norm_pre": 0.06870479521984989,
+      "grad_norm_post": 0.06870479521984989,
+      "learning_rate": 9e-7,
+      "samples_seen": 192
+    },
+    {
+      "step": 4,
+      "loss": 0.03268800624102926,
+      "grad_norm_pre": 0.15994647974291268,
+      "grad_norm_post": 0.15994647974291268,
+      "learning_rate": 1.2e-6,
+      "samples_seen": 256
+    },
+    {
+      "step": 5,
+      "loss": 0.03540466956607878,
+      "grad_norm_pre": 0.07796946867216864,
+      "grad_norm_post": 0.07796946867216864,
+      "learning_rate": 1.4999999999999998e-6,
+      "samples_seen": 320
+    },
+    {
+      "step": 6,
+      "loss": 0.027722052435277388,
+      "grad_norm_pre": 0.11633184139539246,
+      "grad_norm_post": 0.11633184139539246,
+      "learning_rate": 1.8e-6,
+      "samples_seen": 384
+    },
+    {
+      "step": 7,
+      "loss": 0.03432064541904802,
+      "grad_norm_pre": 0.17371458729824793,
+      "grad_norm_post": 0.17371458729824793,
+      "learning_rate": 2.1e-6,
+      "samples_seen": 448
+    },
+    {
+      "step": 8,
+      "loss": 0.0243554438932299,
+      "grad_norm_pre": 0.05178564291500989,
+      "grad_norm_post": 0.05178564291500989,
+      "learning_rate": 2.4e-6,
+      "samples_seen": 512
+    },
+    {
+      "step": 9,
+      "loss": 0.03652638504573583,
+      "grad_norm_pre": 0.1883752418556006,
+      "grad_norm_post": 0.1883752418556006,
+      "learning_rate": 2.6999999999999996e-6,
+      "samples_seen": 576
+    },
+    {
+      "step": 10,
+      "loss": 0.037582458725922085,
+      "grad_norm_pre": 0.18590227536927145,
+      "grad_norm_post": 0.18590227536927145,
+      "learning_rate": 2.9999999999999997e-6,
+      "samples_seen": 640
+    }
+  ]
+}