XPK Start: Thu Apr 16 22:14:41 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-16 22:15:05.369778: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0416 22:15:05.547516 138286539802432 max_utils.py:273] Attempting to initialize the jax distributed system... I0416 22:15:14.587574 138286539802432 distributed.py:149] Starting JAX distributed service on [::]:8482 I0416 22:15:14.589897 138286539802432 distributed.py:172] Connecting to JAX distributed service on mt-02-synthetic-tipl3-slice-job-0-0.mt-02-synthetic-tipl3:8482 I0416 22:15:15.525028 138286539802432 max_utils.py:284] Jax distributed system initialized! I0416 22:15:21.685937 138286539802432 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 22:15:21.686037 138286539802432 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 22:15:21.686076 138286539802432 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0416 22:15:21.686111 138286539802432 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 22:15:22.349915 138286539802432 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 22:15:22.464040 138286539802432 checkpointing.py:688] Setting up checkpoint logger... I0416 22:15:22.464154 138286539802432 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 22:15:22.464200 138286539802432 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0416 22:15:22.464401 138286539802432 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7dc4bc118f80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 22:15:25.421797 138286539802432 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 22:15:25.422053 138286539802432 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db0c00929c0>}, handler_registry=None I0416 22:15:25.422297 138286539802432 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db0c00929c0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0416 22:15:25.422344 138286539802432 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db0a84fce00>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0416 22:15:25.422377 138286539802432 composite_checkpoint_handler.py:505] Initialized registry DefaultCheckpointHandlerRegistry({('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db0c00929c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db0c00929c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db0a84fce00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db0a84fce00>}). I0416 22:15:25.422703 138286539802432 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0416 22:15:25.422777 138286539802432 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7db0c0272020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0416 22:15:26.551947 138286539802432 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints I0416 22:15:26.982309 138286539802432 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7db0c06f8530> I0416 22:15:26.982483 138286539802432 checkpointing.py:302] Checkpoint manager created! I0416 22:15:27.939746 138286539802432 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 22:15:27.939868 138286539802432 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0416 22:15:29.338506 138286539802432 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 22:15:29.338622 138286539802432 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:15:29.338675 138286539802432 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:15:29.338717 138286539802432 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 22:15:29.338767 138286539802432 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:15:29.338809 138286539802432 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:15:29.338915 138286539802432 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:15:29.338975 138286539802432 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 22:15:29.339028 138286539802432 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:15:29.339072 138286539802432 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:15:29.339107 138286539802432 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 22:15:29.339140 138286539802432 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 22:15:29.739520 138286539802432 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:15:29.739612 138286539802432 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:15:29.745167 138286539802432 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:15:29.745220 138286539802432 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:15:29.761577 138286539802432 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:15:29.761644 138286539802432 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:15:29.777103 138286539802432 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:15:29.777157 138286539802432 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:15:29.800982 138286539802432 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:15:29.801046 138286539802432 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:15:29.821767 138286539802432 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:15:29.821876 138286539802432 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:15:29.837518 138286539802432 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:15:29.837582 138286539802432 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:15:29.867703 138286539802432 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 22:15:29.867771 138286539802432 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:15:29.887377 138286539802432 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 22:15:29.887450 138286539802432 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0416 22:15:40.709774 138286539802432 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0416 22:15:40.718584 138286539802432 metric_logger.py:289] number parameters: 1.104 billion I0416 22:15:52.001300 138286539802432 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 22:15:52.152683 138286539802432 checkpointing.py:798] Waited 0.1513671875 seconds for step 0 to finish before starting checkpointing. I0416 22:15:52.154945 138286539802432 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:15:52.156505 138286539802432 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0416 22:15:52.157975 138286539802432 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0. I0416 22:15:52.425799 138286539802432 signaling_client.py:364] Using JaxDistributedSignalingClient I0416 22:15:52.444511 138286539802432 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:15:52.444582 138286539802432 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0416 22:15:52.679540 138158370588416 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0 I0416 22:15:52.726969 138286539802432 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.283818s I0416 22:15:52.727722 138286539802432 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.238 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29447126388549805 s) (per-host) I0416 22:15:52.727796 138286539802432 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.294564s (batch_requests_ready=0.002978s, total_serialization_initiated=0.290910s, others=0.000676s) I0416 22:15:52.727896 138286539802432 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.298607s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.298589) I0416 22:15:52.728630 138286539802432 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.57 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0. I0416 22:15:52.728977 138158378981120 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:35:52.728937 I0416 22:15:52.730968 138286539802432 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:15:52.731287 138156238857984 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:15:52.731441 138286539802432 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776377752.1549282, 'wait_for_prev_duration_secs': 5.698204040527344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776377752.1565416, 'checkpointer_blocking_duration_secs': 0.5725748538970947, 'get_old_steps_start_time': 1776377752.7291415, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776377752.1531038, 'checkpoint_manager_blocking_duration_secs': 0.5782995223999023} I0416 22:15:52.731555 138286539802432 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 22:15:52.731624 138286539802432 max_utils.py:750] Memstats: After params initialized: I0416 22:15:52.731674 138286539802432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0416 22:15:52.731708 138286539802432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0416 22:15:52.731732 138286539802432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0416 22:15:52.731754 138286539802432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0416 22:15:53.069473 138286539802432 metric_logger.py:185] completed step: 0, seconds: 10.989, TFLOP/s/device: 1.236, Tokens/s/device: 186.363, total_weights: 65536, loss: 10.874 I0416 22:15:53.072670 138286539802432 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/tensorboard/' I0416 22:15:53.501561 138286539802432 metric_logger.py:185] completed step: 1, seconds: 1.065, TFLOP/s/device: 12.762, Tokens/s/device: 1923.644, total_weights: 65536, loss: 10.874 I0416 22:15:53.643250 138286539802432 metric_logger.py:185] completed step: 2, seconds: 0.433, TFLOP/s/device: 31.371, Tokens/s/device: 4728.645, total_weights: 65536, loss: 10.262 I0416 22:15:53.790867 138286539802432 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1128.874, Tokens/s/device: 170156.198, total_weights: 65536, loss: 9.732 I0416 22:15:55.027378 138157882337024 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/items I0416 22:15:55.239684 138155686381312 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377754778792534, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776377754778792534, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 22:15:56.770426 2827 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0416 22:15:58.951747 138157343774464 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/items/array_metadatas/process_0 I0416 22:16:16.399964 138286539802432 metric_logger.py:185] completed step: 4, seconds: 0.144, TFLOP/s/device: 94.636, Tokens/s/device: 14264.520, total_weights: 65536, loss: 9.274 I0416 22:16:16.409642 138286539802432 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.525, Tokens/s/device: 13946.394, total_weights: 65536, loss: 8.888 I0416 22:16:16.554302 138286539802432 metric_logger.py:185] completed step: 6, seconds: 22.607, TFLOP/s/device: 0.601, Tokens/s/device: 90.591, total_weights: 65536, loss: 8.589 I0416 22:16:16.702035 138286539802432 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1778.420, Tokens/s/device: 268062.827, total_weights: 65536, loss: 8.380 I0416 22:16:16.849520 138286539802432 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.092, Tokens/s/device: 14031.818, total_weights: 65536, loss: 8.251 I0416 22:16:16.997868 138286539802432 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 22:16:16.998542 138286539802432 checkpointing.py:798] Waited 0.0006947517395019531 seconds for step 9 to finish before starting checkpointing. I0416 22:16:17.000396 138286539802432 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:16:33.520577 138157301913344 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.756087s (commit=35.502159s, array_metadata_write=1.253928s) I0416 22:16:33.521781 138158378981120 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 38.444 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.08848333358765 s) (per-host) I0416 22:16:33.521925 138158378981120 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.792840s. I0416 22:16:34.211153 138158378981120 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377754778792534, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 22:16:34.931683 138158378981120 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000440s. I0416 22:16:35.007605 138155686381312 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377754778792534, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 22:16:35.286088 138158378981120 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:16:36.241075 138158378981120 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.893584s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/items I0416 22:16:36.242782 138158378981120 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0/items I0416 22:16:36.811403 138158378981120 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0 I0416 22:16:38.224790 138158378981120 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0`. I0416 22:16:38.225618 138158378981120 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.07 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/0 I0416 22:16:38.227053 138158378981120 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.497967s. I0416 22:16:38.227261 138156238857984 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:16:38.227374 138156238857984 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:16:38.227439 138156238857984 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:16:38.229634 138156238857984 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 22:16:38.229811 138286539802432 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0416 22:16:38.229959 138286539802432 checkpoint_manager.py:1452] Waiting for previous save to complete took 21.229560 seconds. If this number is high, consider checkpointing less frequently. I0416 22:16:38.231926 138286539802432 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0416 22:16:38.234104 138286539802432 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9. I0416 22:16:38.812216 138286539802432 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:16:38.812306 138286539802432 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0416 22:16:38.839289 138286539802432 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028536s I0416 22:16:38.840078 138286539802432 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.659 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03702855110168457 s) (per-host) I0416 22:16:38.840145 138286539802432 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037114s (batch_requests_ready=0.002597s, total_serialization_initiated=0.033777s, others=0.000740s) I0416 22:16:38.840218 138286539802432 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041563s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.041547) I0416 22:16:38.840925 138286539802432 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9. I0416 22:16:38.841252 138156238857984 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:36:38.841218 I0416 22:16:38.852979 138286539802432 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:16:38.853274 138156808595200 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:16:38.853441 138286539802432 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776377777.000367, 'wait_for_prev_duration_secs': 21.229560375213623, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776377798.2319646, 'checkpointer_blocking_duration_secs': 0.6094205379486084, 'get_old_steps_start_time': 1776377798.8414094, 'get_old_steps_duration_secs': 3.361701965332031e-05, 'checkpoint_manager_blocking_start_time': 1776377776.9987602, 'checkpoint_manager_blocking_duration_secs': 21.854646921157837} I0416 22:16:38.853553 138286539802432 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 22:16:38.853598 138286539802432 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:16:39.038889 138158378981120 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9 I0416 22:16:40.771385 138157873944320 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9/items I0416 22:16:44.451061 138157301913344 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9/items/array_metadatas/process_0 I0416 22:17:25.029275 138156816987904 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.324862s (commit=40.350454s, array_metadata_write=1.974408s) I0416 22:17:25.030504 138156238857984 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.170 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.22742772102356 s) (per-host) I0416 22:17:25.030622 138156238857984 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.189261s. I0416 22:17:26.499649 138156238857984 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000416s. I0416 22:17:26.782376 138156238857984 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:17:27.757382 138156238857984 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.879840s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9/items I0416 22:17:27.759216 138156238857984 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9/items I0416 22:17:28.312163 138156238857984 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9 I0416 22:17:30.099023 138156238857984 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9`. I0416 22:17:30.099873 138156238857984 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.87 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_02_synthetic/checkpoints/9 I0416 22:17:30.101321 138156238857984 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.259959s. I0416 22:17:30.101449 138156808595200 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:17:30.101519 138156808595200 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:17:30.101591 138156808595200 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:17:30.104300 138156808595200 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 22:17:30.104468 138286539802432 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0416 22:17:30.104635 138286539802432 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:17:30.105314 138286539802432 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.164, Tokens/s/device: 13891.904, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 16 22:17:42 UTC 2026 EXIT_CODE=0