XPK Start: Fri Apr 17 04:47:04 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 04:47:28.492018: 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) I0417 04:47:29.025437 138717222905664 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 04:47:38.066300 138717222905664 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 04:47:38.068517 138717222905664 distributed.py:172] Connecting to JAX distributed service on mt-03-dropout-dw2vo-slice-job-0-0.mt-03-dropout-dw2vo:8482 I0417 04:47:39.014124 138717222905664 max_utils.py:284] Jax distributed system initialized! I0417 04:47:45.258615 138717222905664 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 04:47:45.258719 138717222905664 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 04:47:45.258759 138717222905664 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 I0417 04:47:45.258792 138717222905664 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 04:47:45.931132 138717222905664 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:47:46.043685 138717222905664 checkpointing.py:688] Setting up checkpoint logger... I0417 04:47:46.043795 138717222905664 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 04:47:46.043857 138717222905664 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 04:47:46.044059 138717222905664 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 0x7e29103aa6c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 04:47:49.012058 138717222905664 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 04:47:49.012293 138717222905664 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 0x7e1548544320>}, handler_registry=None I0417 04:47:49.012528 138717222905664 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e1548544320>` 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`. I0417 04:47:49.012578 138717222905664 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e15485cfe00>` 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`. I0417 04:47:49.012612 138717222905664 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 0x7e1548544320>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e1548544320>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e15485cfe00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e15485cfe00>}). I0417 04:47:49.012944 138717222905664 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 04:47:49.013015 138717222905664 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e1548481ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 04:47:50.511760 138717222905664 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints I0417 04:47:51.382925 138717222905664 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e1548404c50> I0417 04:47:51.383099 138717222905664 checkpointing.py:302] Checkpoint manager created! I0417 04:47:51.742729 138717222905664 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 04:47:51.742855 138717222905664 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 04:47:53.154099 138717222905664 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 04:47:53.154212 138717222905664 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 04:47:53.154264 138717222905664 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 04:47:53.154309 138717222905664 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 04:47:53.154350 138717222905664 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 04:47:53.154386 138717222905664 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 04:47:53.154423 138717222905664 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:47:53.154459 138717222905664 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 04:47:53.154492 138717222905664 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:47:53.154525 138717222905664 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:47:53.154556 138717222905664 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 04:47:53.154588 138717222905664 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 04:47:53.561385 138717222905664 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:47:53.561477 138717222905664 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 04:47:53.566393 138717222905664 nnx_decoders.py:465] Unknown Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:47:53.566448 138717222905664 nnx_decoders.py:465] Unknown Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 04:47:53.582683 138717222905664 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:47:53.582744 138717222905664 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 04:47:53.598495 138717222905664 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:47:53.598554 138717222905664 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 04:47:53.623567 138717222905664 attentions.py:1154] attentions/query Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:47:53.623633 138717222905664 attentions.py:1154] attentions/query Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 04:47:53.639783 138717222905664 attentions.py:1155] attentions/key Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:47:53.639854 138717222905664 attentions.py:1155] attentions/key Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 04:47:53.655603 138717222905664 attentions.py:1156] attentions/value Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:47:53.655662 138717222905664 attentions.py:1156] attentions/value Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 04:47:53.685957 138717222905664 attentions.py:1197] attentions/out Logical: bfloat16[32,128,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 04:47:53.686029 138717222905664 attentions.py:1197] attentions/out Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 04:47:53.713849 138717222905664 linears.py:525] linears/x Logical: bfloat16[32,128,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 04:47:53.713915 138717222905664 linears.py:525] linears/x Physical: bfloat16[32,128,7168]....................................... ('fsdp', None, None). I0417 04:47:57.415439 138717222905664 max_utils.py:791] Total memory size: 0.8 GB, Output size: 0.4 GB, Temp size: 0.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0417 04:47:57.424591 138717222905664 metric_logger.py:289] number parameters: 1.104 billion I0417 04:48:01.554536 138717222905664 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 04:48:01.628669 138717222905664 checkpointing.py:798] Waited 0.07412266731262207 seconds for step 0 to finish before starting checkpointing. I0417 04:48:01.631123 138717222905664 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:48:01.633264 138717222905664 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 04:48:01.634739 138717222905664 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0. I0417 04:48:02.330688 138717222905664 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 04:48:02.349637 138717222905664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 04:48:02.349706 138717222905664 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 I0417 04:48:02.591808 138590819878656 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0 I0417 04:48:02.625973 138717222905664 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277773s I0417 04:48:02.626765 138717222905664 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.343 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2886922359466553 s) (per-host) I0417 04:48:02.626854 138717222905664 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288783s (batch_requests_ready=0.003073s, total_serialization_initiated=0.285011s, others=0.000699s) I0417 04:48:02.626938 138717222905664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292693s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.292677) I0417 04:48:02.627767 138717222905664 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.99 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0. I0417 04:48:02.628133 138590828271360 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:08:02.628091 I0417 04:48:02.634545 138717222905664 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:48:02.634881 138588704458496 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:48:02.635051 138717222905664 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776401281.6311045, 'wait_for_prev_duration_secs': 6.079673767089844e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776401281.6333025, 'checkpointer_blocking_duration_secs': 0.9949815273284912, 'get_old_steps_start_time': 1776401282.6283085, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776401281.6290836, 'checkpoint_manager_blocking_duration_secs': 1.0059270858764648} I0417 04:48:02.635161 138717222905664 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 04:48:02.635231 138717222905664 max_utils.py:750] Memstats: After params initialized: I0417 04:48:02.635281 138717222905664 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_0(process=0,(0,0,0,0)) I0417 04:48:02.635313 138717222905664 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_1(process=0,(1,0,0,0)) I0417 04:48:02.635340 138717222905664 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_4(process=0,(0,1,0,0)) I0417 04:48:02.635365 138717222905664 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_5(process=0,(1,1,0,0)) I0417 04:48:02.985153 138717222905664 metric_logger.py:185] completed step: 0, seconds: 3.932, TFLOP/s/device: 0.204, Tokens/s/device: 32.551, total_weights: 4096, loss: 10.874 I0417 04:48:02.988233 138717222905664 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/tensorboard/' I0417 04:48:03.424702 138717222905664 metric_logger.py:185] completed step: 1, seconds: 1.427, TFLOP/s/device: 0.561, Tokens/s/device: 89.685, total_weights: 4096, loss: 10.892 I0417 04:48:03.489506 138717222905664 metric_logger.py:185] completed step: 2, seconds: 0.441, TFLOP/s/device: 1.818, Tokens/s/device: 290.516, total_weights: 4096, loss: 9.969 I0417 04:48:03.560036 138717222905664 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 66.336, Tokens/s/device: 10602.170, total_weights: 4096, loss: 9.125 I0417 04:48:05.179331 138590331856640 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/items I0417 04:48:05.315288 138587076245248 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776401284923015539, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776401284923015539, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 04:48:07.147078 2781 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 04:48:09.308147 138588729636608 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0417 04:48:13.997203 138717222905664 metric_logger.py:185] completed step: 4, seconds: 0.066, TFLOP/s/device: 12.165, Tokens/s/device: 1944.284, total_weights: 4096, loss: 8.469 I0417 04:48:14.006938 138717222905664 metric_logger.py:185] completed step: 5, seconds: 0.070, TFLOP/s/device: 11.440, Tokens/s/device: 1828.441, total_weights: 4096, loss: 7.908 I0417 04:48:14.074502 138717222905664 metric_logger.py:185] completed step: 6, seconds: 10.436, TFLOP/s/device: 0.077, Tokens/s/device: 12.266, total_weights: 4096, loss: 7.554 I0417 04:48:14.145103 138717222905664 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 106.599, Tokens/s/device: 17037.136, total_weights: 4096, loss: 7.313 I0417 04:48:14.215543 138717222905664 metric_logger.py:185] completed step: 8, seconds: 0.069, TFLOP/s/device: 11.584, Tokens/s/device: 1851.423, total_weights: 4096, loss: 7.091 I0417 04:48:14.287030 138717222905664 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 04:48:14.287701 138717222905664 checkpointing.py:798] Waited 0.0006814002990722656 seconds for step 9 to finish before starting checkpointing. I0417 04:48:14.289752 138717222905664 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:48:39.817886 138588721243904 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.689412s (commit=31.046024s, array_metadata_write=1.643388s) I0417 04:48:39.819021 138590828271360 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 42.144 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.48092031478882 s) (per-host) I0417 04:48:39.819154 138590828271360 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.190908s. I0417 04:48:40.506179 138590828271360 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776401284923015539, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 04:48:41.616690 138590828271360 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000418s. I0417 04:48:41.712792 138587076245248 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776401284923015539, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 04:48:41.905713 138590828271360 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:48:42.821378 138590828271360 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.189922s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/items I0417 04:48:42.823131 138590828271360 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0/items I0417 04:48:43.344411 138590828271360 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0 I0417 04:48:44.703660 138590828271360 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0`. I0417 04:48:44.704517 138590828271360 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 43.07 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/0 I0417 04:48:44.705765 138590828271360 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 42.077518s. I0417 04:48:44.705984 138588704458496 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:48:44.706112 138588704458496 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:48:44.706186 138588704458496 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:48:44.707986 138588704458496 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 04:48:44.708122 138717222905664 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. W0417 04:48:44.708218 138717222905664 checkpoint_manager.py:1452] Waiting for previous save to complete took 30.418468 seconds. If this number is high, consider checkpointing less frequently. I0417 04:48:44.710143 138717222905664 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 04:48:44.712145 138717222905664 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9. I0417 04:48:45.441029 138717222905664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 04:48:45.441125 138717222905664 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 I0417 04:48:45.475467 138717222905664 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035868s I0417 04:48:45.476294 138717222905664 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 33.822 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0456087589263916 s) (per-host) I0417 04:48:45.476364 138717222905664 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045695s (batch_requests_ready=0.003423s, total_serialization_initiated=0.041528s, others=0.000744s) I0417 04:48:45.476446 138717222905664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049625s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.049609) I0417 04:48:45.477149 138717222905664 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9. I0417 04:48:45.477443 138588704458496 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:08:45.477420 I0417 04:48:45.483146 138717222905664 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:48:45.483385 138589266507520 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:48:45.483510 138717222905664 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776401294.2897193, 'wait_for_prev_duration_secs': 30.418468475341797, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776401324.710182, 'checkpointer_blocking_duration_secs': 0.7673602104187012, 'get_old_steps_start_time': 1776401325.4775653, 'get_old_steps_duration_secs': 3.1948089599609375e-05, 'checkpoint_manager_blocking_start_time': 1776401294.2879562, 'checkpoint_manager_blocking_duration_secs': 31.195520162582397} I0417 04:48:45.483619 138717222905664 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 04:48:45.483663 138717222905664 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:48:45.668874 138590828271360 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9 I0417 04:48:49.569489 138585453094656 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9/items I0417 04:48:52.858794 138589794985728 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0417 04:49:30.918181 138589778200320 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.243610s (commit=39.013809s, array_metadata_write=1.229801s) I0417 04:49:30.919296 138588704458496 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.725 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.48857665061951 s) (per-host) I0417 04:49:30.919410 138588704458496 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.441891s. I0417 04:49:33.241915 138588704458496 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000417s. I0417 04:49:33.508769 138588704458496 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:49:34.438345 138588704458496 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.197139s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9/items I0417 04:49:34.441165 138588704458496 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9/items I0417 04:49:34.946812 138588704458496 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9 I0417 04:49:36.322483 138588704458496 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9`. I0417 04:49:36.323176 138588704458496 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.61 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_03_dropout/checkpoints/9 I0417 04:49:36.324389 138588704458496 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.846848s. I0417 04:49:36.324692 138589266507520 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:49:36.324816 138589266507520 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:49:36.324895 138589266507520 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:49:36.326542 138589266507520 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 04:49:36.326722 138717222905664 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. I0417 04:49:36.326926 138717222905664 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:49:36.327653 138717222905664 metric_logger.py:185] completed step: 9, seconds: 0.070, TFLOP/s/device: 11.411, Tokens/s/device: 1823.752, total_weights: 4096, loss: 7.016 Per train step: Total TFLOPs: 0.80 split as 99.60% learnable weight flops and 0.40% attention flops XPK End: Fri Apr 17 04:49:50 UTC 2026 EXIT_CODE=0