Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+6000/ema_model/ocdbt.process_0/d/326330dfb7358437e8e59be06768480a filter=lfs diff=lfs merge=lfs -text
+6000/model/ocdbt.process_0/d/4227ad9db6fe18c182739366deb973bf filter=lfs diff=lfs merge=lfs -text
+6000/model/ocdbt.process_0/d/d50dd91db86cfae2a2dbddb8642f8d99 filter=lfs diff=lfs merge=lfs -text
+6000/opt/ocdbt.process_0/d/b4f5cc738d2d99797d79e32073d5e69d filter=lfs diff=lfs merge=lfs -text
+6000/opt/ocdbt.process_0/d/c6025d199aab80ff2dbe3a97b9884055 filter=lfs diff=lfs merge=lfs -text

6000/_CHECKPOINT_METADATA

@@ -0,0 +1 @@
+{"item_handlers": {"ema_model": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler", "metrics": "orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler", "model": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler", "opt": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler"}, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1769923330904572953, "commit_timestamp_nsecs": 1769923331516984389, "custom_metadata": {}}

6000/ema_model/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('Q_head', 'layers', '1', 'bias', 'value')": {"key_metadata": [{"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1]}}, "('Q_head', 'layers', '1', 'kernel', 'value')": {"key_metadata": [{"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1]}}, "('init_y', 'scale')": {"key_metadata": [{"key": "init_y", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('init_y', 'state')": {"key_metadata": [{"key": "init_y", "key_type": 2}, {"key": "state", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1, 1, 128]}}, "('init_z', 'scale')": {"key_metadata": [{"key": "init_z", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('init_z', 'state')": {"key_metadata": [{"key": "init_z", "key_type": 2}, {"key": "state", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1, 1, 128]}}, "('input_embedding', 'embedding', 'value')": {"key_metadata": [{"key": "input_embedding", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10, 128]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '2', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'x_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "x_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'y_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "y_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'z_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "z_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('output_head', 'bias', 'value')": {"key_metadata": [{"key": "output_head", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10]}}, "('output_head', 'kernel', 'value')": {"key_metadata": [{"key": "output_head", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 10]}}}, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

6000/ema_model/_sharding

@@ -0,0 +1 @@
+{"UV9oZWFkLmxheWVycy4xLmJpYXMudmFsdWU=":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","UV9oZWFkLmxheWVycy4xLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF95LnN0YXRl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF95LnNjYWxl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF96LnN0YXRl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF96LnNjYWxl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5wdXRfZW1iZWRkaW5nLmVtYmVkZGluZy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3V0cHV0X2hlYWQuYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3V0cHV0X2hlYWQua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMi5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnhfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnlfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnpfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}"}

6000/ema_model/array_metadatas/process_0

@@ -0,0 +1 @@
+{"array_metadatas": [{"array_metadata": {"param_name": "Q_head.layers.1.bias.value", "write_shape": [1], "chunk_shape": [1], "ext_metadata": null}}, {"array_metadata": {"param_name": "Q_head.layers.1.kernel.value", "write_shape": [128, 1], "chunk_shape": [128, 1], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_y.scale", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_y.state", "write_shape": [1, 1, 128], "chunk_shape": [1, 1, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_z.scale", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_z.state", "write_shape": [1, 1, 128], "chunk_shape": [1, 1, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "input_embedding.embedding.value", "write_shape": [10, 128], "chunk_shape": [10, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.2.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.x_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.y_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.z_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "output_head.bias.value", "write_shape": [10], "chunk_shape": [10], "ext_metadata": null}}, {"array_metadata": {"param_name": "output_head.kernel.value", "write_shape": [128, 10], "chunk_shape": [128, 10], "ext_metadata": null}}]}

6000/ema_model/ocdbt.process_0/d/326330dfb7358437e8e59be06768480a

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcf22eff0f560daba7591284cbcf9ac888f376293b6e4493f05986820009f21d
+size 18001920

6000/metrics/metrics

@@ -0,0 +1 @@
+{"eval/cell_acc": 0.915348470211029, "eval/cell_acc_first_delta": 0.20362329483032227, "eval/cell_acc_halfway_delta": 0.010191619396209717, "eval/cell_acc_train_length_delta": 0.013503074645996094, "eval/effective_N_supervision": 32.0, "eval/solved_acc": 0.7916666865348816, "eval/solved_acc_first_delta": 0.703125, "eval/solved_acc_halfway_delta": 0.029947936534881592, "eval/solved_acc_train_length_delta": 0.04036456346511841}

6000/model/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('Q_head', 'layers', '1', 'bias', 'value')": {"key_metadata": [{"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1]}}, "('Q_head', 'layers', '1', 'kernel', 'value')": {"key_metadata": [{"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1]}}, "('init_y', 'scale')": {"key_metadata": [{"key": "init_y", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('init_y', 'state')": {"key_metadata": [{"key": "init_y", "key_type": 2}, {"key": "state", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1, 1, 128]}}, "('init_z', 'scale')": {"key_metadata": [{"key": "init_z", "key_type": 2}, {"key": "scale", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('init_z', 'state')": {"key_metadata": [{"key": "init_z", "key_type": 2}, {"key": "state", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1, 1, 128]}}, "('input_embedding', 'embedding', 'value')": {"key_metadata": [{"key": "input_embedding", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10, 128]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('net', 'net', 'layers', '0', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('net', 'net', 'layers', '0', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('net', 'net', 'layers', '1', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('net', 'net', 'layers', '1', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('net', 'net', 'layers', '2', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'x_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "x_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'y_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "y_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('net', 'z_norm', 'scale', 'value')": {"key_metadata": [{"key": "net", "key_type": 2}, {"key": "z_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('output_head', 'bias', 'value')": {"key_metadata": [{"key": "output_head", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10]}}, "('output_head', 'kernel', 'value')": {"key_metadata": [{"key": "output_head", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 10]}}}, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

6000/model/_sharding

@@ -0,0 +1 @@
+{"UV9oZWFkLmxheWVycy4xLmJpYXMudmFsdWU=":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","UV9oZWFkLmxheWVycy4xLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF95LnN0YXRl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF95LnNjYWxl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF96LnN0YXRl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5pdF96LnNjYWxl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","aW5wdXRfZW1iZWRkaW5nLmVtYmVkZGluZy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3V0cHV0X2hlYWQuYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3V0cHV0X2hlYWQua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5kX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMC5sX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5kX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlcxLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlcyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMS5sX21peGVyLlczLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lm5ldC5sYXllcnMuMi5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnhfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnlfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","bmV0Lnpfbm9ybS5zY2FsZS52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}"}

6000/model/array_metadatas/process_0

@@ -0,0 +1 @@
+{"array_metadatas": [{"array_metadata": {"param_name": "Q_head.layers.1.bias.value", "write_shape": [1], "chunk_shape": [1], "ext_metadata": null}}, {"array_metadata": {"param_name": "Q_head.layers.1.kernel.value", "write_shape": [128, 1], "chunk_shape": [128, 1], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_y.scale", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_y.state", "write_shape": [1, 1, 128], "chunk_shape": [1, 1, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_z.scale", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "init_z.state", "write_shape": [1, 1, 128], "chunk_shape": [1, 1, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "input_embedding.embedding.value", "write_shape": [10, 128], "chunk_shape": [10, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.0.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.1.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.net.layers.2.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.x_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.y_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "net.z_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "output_head.bias.value", "write_shape": [10], "chunk_shape": [10], "ext_metadata": null}}, {"array_metadata": {"param_name": "output_head.kernel.value", "write_shape": [128, 10], "chunk_shape": [128, 10], "ext_metadata": null}}]}

6000/model/ocdbt.process_0/d/4227ad9db6fe18c182739366deb973bf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28e7740ce21a334f62b8a343df51f2f2796fc1593b4cc928322d39ec1e732991
+size 6012928

6000/model/ocdbt.process_0/d/d50dd91db86cfae2a2dbddb8642f8d99

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fa282b093b625ca1f170c154ff4e0b39bf3e91c998f60e16712adf4bcb7f394
+size 12038144

6000/opt/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('opt_state', '1', '0', 'count', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('opt_state', '1', '0', 'mu', 'Q_head', 'layers', '1', 'bias', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1]}}, "('opt_state', '1', '0', 'mu', 'Q_head', 'layers', '1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1]}}, "('opt_state', '1', '0', 'mu', 'input_embedding', 'embedding', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "input_embedding", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10, 128]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '0', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '1', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'mu', 'net', 'net', 'layers', '2', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'mu', 'net', 'x_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "x_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'mu', 'net', 'y_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "y_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'mu', 'net', 'z_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "z_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'mu', 'output_head', 'bias', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "output_head", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10]}}, "('opt_state', '1', '0', 'mu', 'output_head', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "mu", "key_type": 2}, {"key": "output_head", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 10]}}, "('opt_state', '1', '0', 'nu', 'Q_head', 'layers', '1', 'bias', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1]}}, "('opt_state', '1', '0', 'nu', 'Q_head', 'layers', '1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "Q_head", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1]}}, "('opt_state', '1', '0', 'nu', 'input_embedding', 'embedding', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "input_embedding", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10, 128]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '0', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'd_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'd_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [384, 512]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'd_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "d_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 1536]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'l_mixer', 'W1', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W1", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'l_mixer', 'W2', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W2", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [64, 81]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '1', 'l_mixer', 'W3', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "l_mixer", "key_type": 2}, {"key": "W3", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [81, 64]}}, "('opt_state', '1', '0', 'nu', 'net', 'net', 'layers', '2', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'nu', 'net', 'x_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "x_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'nu', 'net', 'y_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "y_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'nu', 'net', 'z_norm', 'scale', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "net", "key_type": 2}, {"key": "z_norm", "key_type": 2}, {"key": "scale", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128]}}, "('opt_state', '1', '0', 'nu', 'output_head', 'bias', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "output_head", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [10]}}, "('opt_state', '1', '0', 'nu', 'output_head', 'kernel', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "nu", "key_type": 2}, {"key": "output_head", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [128, 10]}}, "('opt_state', '1', '2', 'count', 'value')": {"key_metadata": [{"key": "opt_state", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('step', 'value')": {"key_metadata": [{"key": "step", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}}, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

6000/opt/_sharding

@@ -0,0 +1 @@
+{"b3B0X3N0YXRlLjEuMC5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5RX2hlYWQubGF5ZXJzLjEuYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5RX2hlYWQubGF5ZXJzLjEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5pbnB1dF9lbWJlZGRpbmcuZW1iZWRkaW5nLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQubmV0LmxheWVycy4yLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQueF9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQueV9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5uZXQuel9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5vdXRwdXRfaGVhZC5iaWFzLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5tdS5vdXRwdXRfaGVhZC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5RX2hlYWQubGF5ZXJzLjEuYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5RX2hlYWQubGF5ZXJzLjEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5pbnB1dF9lbWJlZGRpbmcuZW1iZWRkaW5nLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmRfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4wLmxfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmRfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzEua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzIua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4xLmxfbWl4ZXIuVzMua2VybmVsLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQubmV0LmxheWVycy4yLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQueF9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQueV9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5uZXQuel9ub3JtLnNjYWxlLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5vdXRwdXRfaGVhZC5iaWFzLnZhbHVl":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMC5udS5vdXRwdXRfaGVhZC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","b3B0X3N0YXRlLjEuMi5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}","c3RlcC52YWx1ZQ==":"{\"sharding_type\": \"NamedSharding\", \"shape\": [4], \"axis_names\": [\"data\"], \"axis_types\": [\"AxisType.Auto\"], \"partition_spec\": [], \"device_mesh\": {\"mesh\": [{\"id\": 0}, {\"id\": 2}, {\"id\": 1}, {\"id\": 3}]}}"}

6000/opt/array_metadatas/process_0

@@ -0,0 +1 @@
+{"array_metadatas": [{"array_metadata": {"param_name": "opt_state.1.0.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.Q_head.layers.1.bias.value", "write_shape": [1], "chunk_shape": [1], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.Q_head.layers.1.kernel.value", "write_shape": [128, 1], "chunk_shape": [128, 1], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.input_embedding.embedding.value", "write_shape": [10, 128], "chunk_shape": [10, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.0.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.1.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.net.layers.2.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.x_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.y_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.net.z_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.output_head.bias.value", "write_shape": [10], "chunk_shape": [10], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.mu.output_head.kernel.value", "write_shape": [128, 10], "chunk_shape": [128, 10], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.Q_head.layers.1.bias.value", "write_shape": [1], "chunk_shape": [1], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.Q_head.layers.1.kernel.value", "write_shape": [128, 1], "chunk_shape": [128, 1], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.input_embedding.embedding.value", "write_shape": [10, 128], "chunk_shape": [10, 128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.0.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.d_mixer.W1.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.d_mixer.W2.kernel.value", "write_shape": [384, 512], "chunk_shape": [384, 512], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.d_mixer.W3.kernel.value", "write_shape": [128, 1536], "chunk_shape": [128, 1536], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.l_mixer.W1.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.l_mixer.W2.kernel.value", "write_shape": [64, 81], "chunk_shape": [64, 81], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.1.l_mixer.W3.kernel.value", "write_shape": [81, 64], "chunk_shape": [81, 64], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.net.layers.2.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.x_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.y_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.net.z_norm.scale.value", "write_shape": [128], "chunk_shape": [128], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.output_head.bias.value", "write_shape": [10], "chunk_shape": [10], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.0.nu.output_head.kernel.value", "write_shape": [128, 10], "chunk_shape": [128, 10], "ext_metadata": null}}, {"array_metadata": {"param_name": "opt_state.1.2.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "step.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}]}

6000/opt/ocdbt.process_0/d/b4f5cc738d2d99797d79e32073d5e69d

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a1a06ad7c5ed905cb5f0973f4350dbef82849f227e9e5bb9de8b2314247a1ee
+size 3748169

6000/opt/ocdbt.process_0/d/c6025d199aab80ff2dbe3a97b9884055

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c663e3d76b1507face856d12a32e3760141ad51b8bfd55fcc3b7779ca44684b7
+size 32145408

events.out.tfevents.1769913372.t1v-n-9c0e4925-w-0.350823.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42f9304915c63fd98b46d82c37380f49389f783809b2195148d564995530a0d8
+size 24752080

events.out.tfevents.1770133081.t1v-n-9c0e4925-w-0.1844962.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49b309dd9c011c4f7ca6f9a4c8ee77e63a5670190aa20c45e1854f6a72789da3
+size 1720

events.out.tfevents.1770133173.t1v-n-9c0e4925-w-0.1849917.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f52565dd4620d3ed16e57fdea0f6d5be507e3bc63d4565e1cab767ef391a9dc3
+size 1898

events.out.tfevents.1770225197.t1v-n-9c0e4925-w-0.86059.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f54dfef4a22363b5aed269320849ddb5e7ca964efa52d0b51ce7185f189a736
+size 42092

events.out.tfevents.1770266101.t1v-n-9c0e4925-w-0.1112126.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b830f905e140a41307dae897999a967005ba77393016082813d4bcea946d9731
+size 1164

plugins/profile/2026_02_01_02_37_10/t1v-n-9c0e4925-w-0.trace.json.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:632d32d68c4760f7590498eead116a24593898100f4604acebffed55b0ed9090
+size 27903561

plugins/profile/2026_02_01_02_37_10/t1v-n-9c0e4925-w-0.xplane.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd4bcbdfed640d60aa800a9dce9a67153a90ab282af3688ce3e958ff3fffc81c
+size 188857492

plugins/profile/2026_02_01_03_37_09/t1v-n-9c0e4925-w-0.trace.json.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ced01432f4105c4f496ee6c0f648b54a0ad0bf645f1fbff665702f9a5fe9e200
+size 27568933

plugins/profile/2026_02_01_03_37_09/t1v-n-9c0e4925-w-0.xplane.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfa36c5fdd65dd98b0490ce06f007796ba51e176fffdbddadf42a4ba9dd3a04e
+size 188723905

plugins/profile/2026_02_01_04_37_10/t1v-n-9c0e4925-w-0.trace.json.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b26e819bdb02bd36b27c974634fab52031f1611433fb923b216d7823c3f9ca
+size 27701677

plugins/profile/2026_02_01_04_37_10/t1v-n-9c0e4925-w-0.xplane.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c816fdca11b1c33bfbefe8e37b51759a98329cc3060b60e973aa3ea5dec4cb3
+size 188664569

plugins/profile/2026_02_01_05_37_10/t1v-n-9c0e4925-w-0.trace.json.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f57bd0eb47ebaa473ef37ecf15d22672ff350ff23b8337bbbc31af88616d4835
+size 27815077

plugins/profile/2026_02_01_05_37_10/t1v-n-9c0e4925-w-0.xplane.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f347efc7c50d08448814ebff623f78138529802859917ed4d3e66228f468f55d
+size 188718113