Add MiniLLM Orbax checkpoint

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+small_checkpoint.orbax/ocdbt.process_0/d/31f8327a635fa6934c92766cdfcd4ea6 filter=lfs diff=lfs merge=lfs -text
+small_checkpoint.orbax/ocdbt.process_0/d/a1bcbb54b4a62c4936e7676087a47f21 filter=lfs diff=lfs merge=lfs -text
+small_checkpoint.orbax/ocdbt.process_0/d/d07fe683022e705c64989db899c3c8a5 filter=lfs diff=lfs merge=lfs -text

small_checkpoint.orbax/_CHECKPOINT_METADATA

@@ -0,0 +1 @@
+{"item_handlers": "orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler", "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1781515629373263111, "commit_timestamp_nsecs": 1781515630115058716, "custom_metadata": {}}

small_checkpoint.orbax/_METADATA

@@ -0,0 +1 @@
+{"tree_metadata": {"('embedding', 'pos_emb', 'embedding', 'value')": {"key_metadata": [{"key": "embedding", "key_type": 2}, {"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": [128, 192]}}, "('embedding', 'token_emb', 'embedding', 'value')": {"key_metadata": [{"key": "embedding", "key_type": 2}, {"key": "token_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, 192]}}, "('output_layer', 'kernel', 'value')": {"key_metadata": [{"key": "output_layer", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 50257]}}, "('transformer_blocks', '0', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '0', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '0', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '0', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '0', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '0', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '0', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '0', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "0", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '1', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '1', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '1', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '1', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '1', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '1', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '1', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '1', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "1", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '2', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '2', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '2', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '2', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '2', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '2', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '2', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '2', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "2", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '3', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '3', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '3', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '3', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '3', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '3', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '3', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '3', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "3", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '4', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '4', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '4', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '4', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '4', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '4', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '4', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '4', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "4", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '5', 'attention', 'key', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '5', 'attention', 'key', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "key", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '5', 'attention', 'out', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192]}}, "('transformer_blocks', '5', 'attention', 'out', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "out", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32, 192]}}, "('transformer_blocks', '5', 'attention', 'query', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '5', 'attention', 'query', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "query", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}, "('transformer_blocks', '5', 'attention', 'value', 'bias', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "bias", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [6, 32]}}, "('transformer_blocks', '5', 'attention', 'value', 'kernel', 'value')": {"key_metadata": [{"key": "transformer_blocks", "key_type": 2}, {"key": "5", "key_type": 2}, {"key": "attention", "key_type": 2}, {"key": "value", "key_type": 2}, {"key": "kernel", "key_type": 2}, {"key": "value", "key_type": 2}], "value_metadata": {"value_type": "jax.Array", "skip_deserialize": false, "write_shape": [192, 6, 32]}}}, "use_ocdbt": true, "use_zarr3": false, "store_array_data_equal_to_fill_value": true, "custom_metadata": null}

small_checkpoint.orbax/_sharding

@@ -0,0 +1 @@
+{"ZW1iZWRkaW5nLnBvc19lbWIuZW1iZWRkaW5nLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","ZW1iZWRkaW5nLnRva2VuX2VtYi5lbWJlZGRpbmcudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","b3V0cHV0X2xheWVyLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjAuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjEuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjIuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjMuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjQuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLm91dC5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLm91dC5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLmtleS5iaWFzLnZhbHVl":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLmtleS5rZXJuZWwudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLnF1ZXJ5LmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLnF1ZXJ5Lmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLnZhbHVlLmJpYXMudmFsdWU=":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}","dHJhbnNmb3JtZXJfYmxvY2tzLjUuYXR0ZW50aW9uLnZhbHVlLmtlcm5lbC52YWx1ZQ==":"{\"sharding_type\": \"SingleDeviceSharding\", \"device_str\": \"cuda:0\"}"}

small_checkpoint.orbax/array_metadatas/process_0

@@ -0,0 +1 @@
+{"array_metadatas": [{"array_metadata": {"param_name": "embedding.pos_emb.embedding.value", "write_shape": [128, 192], "chunk_shape": [128, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "embedding.token_emb.embedding.value", "write_shape": [50257, 192], "chunk_shape": [50257, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "output_layer.kernel.value", "write_shape": [192, 50257], "chunk_shape": [192, 50257], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.0.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.1.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.2.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.3.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.4.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.key.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.key.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.out.bias.value", "write_shape": [192], "chunk_shape": [192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.out.kernel.value", "write_shape": [6, 32, 192], "chunk_shape": [6, 32, 192], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.query.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.query.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.value.bias.value", "write_shape": [6, 32], "chunk_shape": [6, 32], "ext_metadata": null}}, {"array_metadata": {"param_name": "transformer_blocks.5.attention.value.kernel.value", "write_shape": [192, 6, 32], "chunk_shape": [192, 6, 32], "ext_metadata": null}}]}

small_checkpoint.orbax/ocdbt.process_0/d/31f8327a635fa6934c92766cdfcd4ea6

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9afbd007d00955c6da9bde31d854148bf09e738932ebc7e9f7c3789e07400da8
+size 38322176

small_checkpoint.orbax/ocdbt.process_0/d/a1bcbb54b4a62c4936e7676087a47f21

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c863898b2c430d2375ed141555498407d197723774c74c74ee31f88c8d5c3abc
+size 36069376

small_checkpoint.orbax/ocdbt.process_0/d/d07fe683022e705c64989db899c3c8a5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:194db31993f2cceab658d8b1219166d2d28eabd930e9a99cedbbf558e2e158ea
+size 137719