XPK Start: Thu Apr 16 22:18:53 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 22:19:17.676443: 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 22:19:18.208648 136592325007168 max_utils.py:273] Attempting to initialize the jax distributed system... I0416 22:19:27.248627 136592325007168 distributed.py:149] Starting JAX distributed service on [::]:8482 I0416 22:19:27.251047 136592325007168 distributed.py:172] Connecting to JAX distributed service on mt-03-dropout-ni8sl-slice-job-0-0.mt-03-dropout-ni8sl:8482 I0416 22:19:28.342331 136592325007168 max_utils.py:284] Jax distributed system initialized! I0416 22:19:33.449455 136592325007168 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 22:19:33.449558 136592325007168 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 22:19:33.449596 136592325007168 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 22:19:33.449629 136592325007168 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 22:19:34.120009 136592325007168 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 22:19:34.232487 136592325007168 checkpointing.py:688] Setting up checkpoint logger... I0416 22:19:34.232599 136592325007168 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 22:19:34.232643 136592325007168 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0416 22:19:34.232859 136592325007168 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 0x7c3a4d11cc50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 22:19:37.114978 136592325007168 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 22:19:37.115221 136592325007168 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 0x7c266c12a180>}, handler_registry=None I0416 22:19:37.115496 136592325007168 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c266c12a180>` 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 22:19:37.115551 136592325007168 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c266c119cd0>` 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 22:19:37.115598 136592325007168 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 0x7c266c12a180>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c266c12a180>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c266c119cd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c266c119cd0>}). I0416 22:19:37.115980 136592325007168 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0416 22:19:37.116065 136592325007168 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c266c1d9ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0416 22:19:38.217997 136592325007168 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_03_dropout/checkpoints I0416 22:19:38.235089 136592325007168 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_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c266c13b860> I0416 22:19:38.235213 136592325007168 checkpointing.py:302] Checkpoint manager created! I0416 22:19:38.595066 136592325007168 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 22:19:38.595187 136592325007168 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0416 22:19:39.974456 136592325007168 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 22:19:39.974571 136592325007168 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:19:39.974623 136592325007168 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:19:39.974688 136592325007168 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 22:19:39.974730 136592325007168 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:19:39.974765 136592325007168 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:19:39.974800 136592325007168 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:19:39.974854 136592325007168 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 22:19:39.974892 136592325007168 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:19:39.974924 136592325007168 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:19:39.974955 136592325007168 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 22:19:39.974986 136592325007168 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 22:19:40.380747 136592325007168 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:19:40.380855 136592325007168 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0416 22:19:40.385798 136592325007168 nnx_decoders.py:465] Unknown Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:19:40.385870 136592325007168 nnx_decoders.py:465] Unknown Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0416 22:19:40.401898 136592325007168 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:19:40.401955 136592325007168 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0416 22:19:40.417474 136592325007168 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:19:40.417532 136592325007168 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0416 22:19:40.441492 136592325007168 attentions.py:1154] attentions/query Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:19:40.441556 136592325007168 attentions.py:1154] attentions/query Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0416 22:19:40.457131 136592325007168 attentions.py:1155] attentions/key Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:19:40.457196 136592325007168 attentions.py:1155] attentions/key Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0416 22:19:40.472659 136592325007168 attentions.py:1156] attentions/value Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:19:40.472720 136592325007168 attentions.py:1156] attentions/value Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0416 22:19:40.505564 136592325007168 attentions.py:1197] attentions/out Logical: bfloat16[32,128,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 22:19:40.505638 136592325007168 attentions.py:1197] attentions/out Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0416 22:19:40.531606 136592325007168 linears.py:525] linears/x Logical: bfloat16[32,128,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 22:19:40.531671 136592325007168 linears.py:525] linears/x Physical: bfloat16[32,128,7168]....................................... ('fsdp', None, None). I0416 22:19:44.138602 136592325007168 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. I0416 22:19:44.147823 136592325007168 metric_logger.py:289] number parameters: 1.104 billion I0416 22:19:48.299860 136592325007168 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 22:19:48.372679 136592325007168 checkpointing.py:798] Waited 0.07281756401062012 seconds for step 0 to finish before starting checkpointing. I0416 22:19:48.375118 136592325007168 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:19:48.377168 136592325007168 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0416 22:19:48.378660 136592325007168 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_03_dropout/checkpoints/0. I0416 22:19:49.075035 136592325007168 signaling_client.py:364] Using JaxDistributedSignalingClient I0416 22:19:49.094653 136592325007168 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:19:49.094725 136592325007168 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 22:19:49.338993 136467368089344 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_03_dropout/checkpoints/0 I0416 22:19:49.362702 136592325007168 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269480s I0416 22:19:49.363652 136592325007168 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.492 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2808825969696045 s) (per-host) I0416 22:19:49.363729 136592325007168 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280982s (batch_requests_ready=0.003073s, total_serialization_initiated=0.277055s, others=0.000854s) I0416 22:19:49.363816 136592325007168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.284861s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.284844) I0416 22:19:49.364706 136592325007168 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_03_dropout/checkpoints/0. I0416 22:19:49.365063 136467376482048 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:39:49.365024 I0416 22:19:49.379152 136592325007168 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:19:49.379422 136466290161408 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:19:49.379575 136592325007168 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_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776377988.3750997, 'wait_for_prev_duration_secs': 6.079673767089844e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776377988.3772082, 'checkpointer_blocking_duration_secs': 0.9880039691925049, 'get_old_steps_start_time': 1776377989.365238, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776377988.3731515, 'checkpoint_manager_blocking_duration_secs': 1.0063841342926025} I0416 22:19:49.379680 136592325007168 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 22:19:49.379745 136592325007168 max_utils.py:750] Memstats: After params initialized: I0416 22:19:49.379795 136592325007168 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_0(process=0,(0,0,0,0)) I0416 22:19:49.379833 136592325007168 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_1(process=0,(1,0,0,0)) I0416 22:19:49.379875 136592325007168 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_4(process=0,(0,1,0,0)) I0416 22:19:49.379901 136592325007168 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_5(process=0,(1,1,0,0)) I0416 22:19:49.726055 136592325007168 metric_logger.py:185] completed step: 0, seconds: 3.955, TFLOP/s/device: 0.202, Tokens/s/device: 32.364, total_weights: 4096, loss: 10.874 I0416 22:19:49.728927 136592325007168 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_03_dropout/tensorboard/' I0416 22:19:50.151866 136592325007168 metric_logger.py:185] completed step: 1, seconds: 1.423, TFLOP/s/device: 0.563, Tokens/s/device: 89.967, total_weights: 4096, loss: 10.892 I0416 22:19:50.216789 136592325007168 metric_logger.py:185] completed step: 2, seconds: 0.427, TFLOP/s/device: 1.876, Tokens/s/device: 299.861, total_weights: 4096, loss: 9.969 I0416 22:19:50.287184 136592325007168 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 67.825, Tokens/s/device: 10840.108, total_weights: 4096, loss: 9.125 I0416 22:19:51.083873 136466873181952 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_03_dropout/checkpoints/0/items I0416 22:19:51.250283 136464681056000 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377990826437359, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776377990826437359, "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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 22:19:52.194965 2738 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0416 22:19:53.915584 136466344703744 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_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0416 22:20:00.773386 136592325007168 metric_logger.py:185] completed step: 4, seconds: 0.068, TFLOP/s/device: 11.844, Tokens/s/device: 1892.903, total_weights: 4096, loss: 8.469 I0416 22:20:00.783519 136592325007168 metric_logger.py:185] completed step: 5, seconds: 0.068, TFLOP/s/device: 11.711, Tokens/s/device: 1871.701, total_weights: 4096, loss: 7.908 I0416 22:20:00.850878 136592325007168 metric_logger.py:185] completed step: 6, seconds: 10.485, TFLOP/s/device: 0.076, Tokens/s/device: 12.208, total_weights: 4096, loss: 7.554 I0416 22:20:00.921376 136592325007168 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 104.376, Tokens/s/device: 16681.871, total_weights: 4096, loss: 7.313 I0416 22:20:00.991932 136592325007168 metric_logger.py:185] completed step: 8, seconds: 0.069, TFLOP/s/device: 11.615, Tokens/s/device: 1856.310, total_weights: 4096, loss: 7.091 I0416 22:20:01.063477 136592325007168 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 22:20:01.064245 136592325007168 checkpointing.py:798] Waited 0.0007753372192382812 seconds for step 9 to finish before starting checkpointing. I0416 22:20:01.066179 136592325007168 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:20:24.502017 136466298554112 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.314720s (commit=31.134218s, array_metadata_write=1.180502s) I0416 22:20:24.503099 136467376482048 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 44.596 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 35.42030382156372 s) (per-host) I0416 22:20:24.503229 136467376482048 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.138056s. I0416 22:20:25.658332 136467376482048 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377990826437359, '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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 22:20:26.798815 136464681056000 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377990826437359, '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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 22:20:26.812591 136467376482048 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000448s. I0416 22:20:27.026276 136467376482048 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:20:27.958918 136467376482048 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.171221s. 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_03_dropout/checkpoints/0/items I0416 22:20:27.960430 136467376482048 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_03_dropout/checkpoints/0/items I0416 22:20:28.544629 136467376482048 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_03_dropout/checkpoints/0 I0416 22:20:29.939939 136467376482048 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_03_dropout/checkpoints/0`. I0416 22:20:29.940788 136467376482048 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 41.56 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_03_dropout/checkpoints/0 I0416 22:20:29.942219 136467376482048 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 40.577043s. I0416 22:20:29.942409 136466290161408 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:20:29.942520 136466290161408 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:20:29.942585 136466290161408 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:20:29.945374 136466290161408 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 22:20:29.945583 136592325007168 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 22:20:29.945726 136592325007168 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.879545 seconds. If this number is high, consider checkpointing less frequently. I0416 22:20:29.947714 136592325007168 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0416 22:20:29.949728 136592325007168 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_03_dropout/checkpoints/9. I0416 22:20:30.658441 136592325007168 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:20:30.658535 136592325007168 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 22:20:30.692322 136592325007168 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035259s I0416 22:20:30.693006 136592325007168 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 35.043 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.044019222259521484 s) (per-host) I0416 22:20:30.693068 136592325007168 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044098s (batch_requests_ready=0.002722s, total_serialization_initiated=0.040747s, others=0.000628s) I0416 22:20:30.693140 136592325007168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.048088s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.048070) I0416 22:20:30.693805 136592325007168 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 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_03_dropout/checkpoints/9. I0416 22:20:30.694095 136466290161408 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:40:30.694062 I0416 22:20:30.698606 136592325007168 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:20:30.698905 136465799440128 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:20:30.699067 136592325007168 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_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776378001.0661469, 'wait_for_prev_duration_secs': 28.879544973373413, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776378029.9477525, 'checkpointer_blocking_duration_secs': 0.7464699745178223, 'get_old_steps_start_time': 1776378030.6942468, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776378001.0645478, 'checkpoint_manager_blocking_duration_secs': 29.63448429107666} I0416 22:20:30.699175 136592325007168 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 22:20:30.699221 136592325007168 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:20:30.907450 136467376482048 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_03_dropout/checkpoints/9 I0416 22:20:33.084454 136466839611136 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_03_dropout/checkpoints/9/items I0416 22:20:36.207197 136466302740224 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_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0416 22:21:14.255192 136465807832832 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.068782s (commit=38.744281s, array_metadata_write=1.324500s) I0416 22:21:14.256976 136466290161408 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 36.223 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.60795307159424 s) (per-host) I0416 22:21:14.257100 136466290161408 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.562900s. I0416 22:21:15.673504 136466290161408 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000414s. I0416 22:21:16.330992 136466290161408 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:21:17.072789 136466290161408 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.972010s. 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_03_dropout/checkpoints/9/items I0416 22:21:17.074467 136466290161408 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_03_dropout/checkpoints/9/items I0416 22:21:17.640350 136466290161408 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_03_dropout/checkpoints/9 I0416 22:21:19.525585 136466290161408 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_03_dropout/checkpoints/9`. I0416 22:21:19.526405 136466290161408 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.58 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_03_dropout/checkpoints/9 I0416 22:21:19.528549 136466290161408 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.834346s. I0416 22:21:19.528744 136465799440128 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:21:19.528873 136465799440128 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:21:19.528943 136465799440128 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:21:19.530540 136465799440128 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 22:21:19.530706 136592325007168 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 22:21:19.530899 136592325007168 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:21:19.531555 136592325007168 metric_logger.py:185] completed step: 9, seconds: 0.070, TFLOP/s/device: 11.410, Tokens/s/device: 1823.544, 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: Thu Apr 16 22:21:29 UTC 2026 EXIT_CODE=0