gradling push experiments/gpt2/baseline/runs/0002

.gitattributes

@@ -57,3 +57,8 @@ experiments/gpt2/baseline/runs/0001/checkpoints/final/ocdbt.process_0/d/1bbc49fd
 experiments/gpt2/baseline/runs/0001/checkpoints/final/ocdbt.process_0/d/57b03944839ad147aef22c1b80dfe2a3 filter=lfs diff=lfs merge=lfs -text
 experiments/gpt2/baseline/runs/0001/checkpoints/final/ocdbt.process_0/d/96ebaa04bfc4c571ba58b11c35cd3c53 filter=lfs diff=lfs merge=lfs -text
 experiments/gpt2/baseline/runs/0001/checkpoints/final/ocdbt.process_0/d/c97574b954630bea35d0e96c69caac88 filter=lfs diff=lfs merge=lfs -text
+experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/3bea35be1aeaf3d77ce254f1cd4b041d filter=lfs diff=lfs merge=lfs -text
+experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/65a3b2fde94576222d0ce1cf3867e0a5 filter=lfs diff=lfs merge=lfs -text
+experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/abee4b1cce2c719876e5eaa3976620b0 filter=lfs diff=lfs merge=lfs -text
+experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/cafeea56906b8a68679c9f376c14aaf4 filter=lfs diff=lfs merge=lfs -text
+experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/db3067a195dd115c7cca49d37b47253e filter=lfs diff=lfs merge=lfs -text

experiments/gpt2/baseline/runs/0002/checkpoints/final/_CHECKPOINT_METADATA

@@ -0,0 +1 @@
+{"item_handlers": "orbax.checkpoint._src.handlers.standard_checkpoint_handler.StandardCheckpointHandler", "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1774710533763319852, "commit_timestamp_nsecs": 1774710535996671022, "custom_metadata": {}}

experiments/gpt2/baseline/runs/0002/checkpoints/final/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('blocks', 'layers', '0', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '0', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '0', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '0', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '0', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '0', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '0', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '0', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '0', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '0', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '0', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '0', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '1', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '1', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '1', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '1', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '1', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '1', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '1', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '1', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '1', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '1', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '1', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '1', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '2', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '2', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '2', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '2', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '2', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '2', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '2', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '2', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '2', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '2', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '2', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '2', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '3', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '3', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '3', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '3', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '3', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '3', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '3', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '3', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '3', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '3', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '3', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '3', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '4', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '4', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '4', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '4', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '4', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '4', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '4', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '4', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '4', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '4', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '4', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '4', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '5', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '5', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '5', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '5', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '5', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '5', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '5', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '5', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '5', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '5', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '5', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '5', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '6', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '6', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '6', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '6', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '6', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '6', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '6', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '6', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '6', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '6', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '6', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '6', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "6", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '7', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '7', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '7', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '7', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '7', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '7', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '7', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '7', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '7', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '7', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '7', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '7', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "7", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '8', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '8', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '8', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '8', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '8', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '8', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '8', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '8', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '8', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '8', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '8', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '8', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "8", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '9', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '9', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '9', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '9', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '9', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '9', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '9', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '9', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '9', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '9', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '9', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '9', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "9", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '10', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '10', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '10', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '10', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '10', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '10', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '10', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '10', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '10', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '10', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '10', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '10', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "10", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '11', 'ff', 'dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '11', 'ff', 'dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '11', 'ff', 'fc', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072]}}, "('blocks', 'layers', '11', 'ff', 'fc', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "fc", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 3072]}}, "('blocks', 'layers', '11', 'ff', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'ff', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ff", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [3072, 768]}}, "('blocks', 'layers', '11', 'ln1', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'ln1', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ln1", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'ln2', 'beta', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'ln2', 'gamma', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "ln2", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'sa', 'a_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '11', 'sa', 'a_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "a_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('blocks', 'layers', '11', 'sa', 'attn', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "attn", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 2304]}}, "('blocks', 'layers', '11', 'sa', 'proj', 'bias', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('blocks', 'layers', '11', 'sa', 'proj', 'kernel', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "proj", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768, 768]}}, "('blocks', 'layers', '11', 'sa', 'r_dropout', 'rngs', 'count', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "count", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": []}}, "('blocks', 'layers', '11', 'sa', 'r_dropout', 'rngs', 'key', 'value')": {"key_metadata": [{"key": "blocks", "key_type": 2}, {"key": "layers", "key_type": 2}, {"key": "11", "key_type": 2}, {"key": "sa", "key_type": 2}, {"key": "r_dropout", "key_type": 2}, {"key": "rngs", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [2]}}, "('ln_f', 'beta', 'value')": {"key_metadata": [{"key": "ln_f", "key_type": 2}, {"key": "beta", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('ln_f', 'gamma', 'value')": {"key_metadata": [{"key": "ln_f", "key_type": 2}, {"key": "gamma", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [768]}}, "('pos_emb', 'embedding', 'value')": {"key_metadata": [{"key": "pos_emb", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [1024, 768]}}, "('tok_emb', 'embedding', 'value')": {"key_metadata": [{"key": "tok_emb", "key_type": 2}, {"key": "embedding", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [50257, 768]}}}, "use_ocdbt": true, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

experiments/gpt2/baseline/runs/0002/checkpoints/final/_sharding

@@ -0,0 +1 @@
+{"YmxvY2tzLmxheWVycy40LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy40LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy41LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy42LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy43LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy44LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy45LnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4wLnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xLnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5kcm9wb3V0LnJuZ3MuY291bnQudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5kcm9wb3V0LnJuZ3Mua2V5LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5mYy5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5mYy5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5wcm9qLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5mZi5wcm9qLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5sbjEuYmV0YS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5sbjEuZ2FtbWEudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5sbjIuYmV0YS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5sbjIuZ2FtbWEudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5hX2Ryb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5hX2Ryb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5hdHRuLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5wcm9qLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5wcm9qLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5yX2Ryb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMC5zYS5yX2Ryb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5kcm9wb3V0LnJuZ3MuY291bnQudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5kcm9wb3V0LnJuZ3Mua2V5LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5mYy5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5mYy5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5wcm9qLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5mZi5wcm9qLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5sbjEuYmV0YS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5sbjEuZ2FtbWEudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5sbjIuYmV0YS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5sbjIuZ2FtbWEudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5hX2Ryb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5hX2Ryb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5hdHRuLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5wcm9qLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5wcm9qLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5yX2Ryb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4xMS5zYS5yX2Ryb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4yLnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLmRyb3BvdXQucm5ncy5jb3VudC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLmRyb3BvdXQucm5ncy5rZXkudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLmZjLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLmZjLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmZmLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmxuMS5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmxuMS5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmxuMi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLmxuMi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLmF0dG4ua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLmFfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLmFfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLnByb2ouYmlhcy52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLnByb2oua2VybmVsLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLnJfZHJvcG91dC5ybmdzLmNvdW50LnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","YmxvY2tzLmxheWVycy4zLnNhLnJfZHJvcG91dC5ybmdzLmtleS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","bG5fZi5iZXRhLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","bG5fZi5nYW1tYS52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","cG9zX2VtYi5lbWJlZGRpbmcudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dG9rX2VtYi5lbWJlZGRpbmcudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}"}

experiments/gpt2/baseline/runs/0002/checkpoints/final/array_metadatas/process_0

@@ -0,0 +1 @@
+{"array_metadatas": [{"array_metadata": {"param_name": "blocks.layers.0.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.0.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.0.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.1.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.2.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.3.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.4.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.5.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.6.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.7.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.8.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.9.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.10.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.fc.bias.value", "write_shape": [3072], "chunk_shape": [3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.fc.kernel.value", "write_shape": [768, 3072], "chunk_shape": [768, 3072], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ff.proj.kernel.value", "write_shape": [3072, 768], "chunk_shape": [3072, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ln1.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ln1.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ln2.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.ln2.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.a_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.a_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.attn.kernel.value", "write_shape": [768, 2304], "chunk_shape": [768, 2304], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.proj.bias.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.proj.kernel.value", "write_shape": [768, 768], "chunk_shape": [768, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.r_dropout.rngs.count.value", "write_shape": [], "chunk_shape": [], "ext_metadata": null}}, {"array_metadata": {"param_name": "blocks.layers.11.sa.r_dropout.rngs.key.value", "write_shape": [2], "chunk_shape": [2], "ext_metadata": {"random_key_impl": "threefry2x32"}}}, {"array_metadata": {"param_name": "ln_f.beta.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "ln_f.gamma.value", "write_shape": [768], "chunk_shape": [768], "ext_metadata": null}}, {"array_metadata": {"param_name": "pos_emb.embedding.value", "write_shape": [1024, 768], "chunk_shape": [1024, 768], "ext_metadata": null}}, {"array_metadata": {"param_name": "tok_emb.embedding.value", "write_shape": [50257, 768], "chunk_shape": [50257, 768], "ext_metadata": null}}]}

experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/3bea35be1aeaf3d77ce254f1cd4b041d

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da6f40d717a9e9ff413edbecaaef4a50333a4167c557b4a244290fff45437049
+size 102764544

experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/65a3b2fde94576222d0ce1cf3867e0a5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:395eebddebf79ea0d416be4c278318c2a75b3f3efec9aec5ac9d0c081eceb7f5
+size 24125440

experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/abee4b1cce2c719876e5eaa3976620b0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56c766996d02e38c267eb2be087c64724b118e7f860ceb23af5a9d981e987183
+size 2189937

experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/cafeea56906b8a68679c9f376c14aaf4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0a6e0bb88ea222d77776e808d3d3fd45f0741da5e743be1ee26e011168e0e96
+size 189112320

experiments/gpt2/baseline/runs/0002/checkpoints/final/ocdbt.process_0/d/db3067a195dd115c7cca49d37b47253e

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb05f1bdd15fe1c7ecdf7abf51bf990e1d800f64468c845541f9dc78f4d718ea
+size 142778368