XPK Start: Fri Apr 17 00:43:21 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 00:43:45.114463: 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 00:43:45.290030 140406029731648 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 00:43:54.330954 140406029731648 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 00:43:54.333336 140406029731648 distributed.py:172] Connecting to JAX distributed service on mt-03-dropout-0b5sw-slice-job-0-0.mt-03-dropout-0b5sw:8482 I0417 00:43:55.244300 140406029731648 max_utils.py:284] Jax distributed system initialized! I0417 00:44:01.251293 140406029731648 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 00:44:01.251392 140406029731648 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 00:44:01.251431 140406029731648 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 00:44:01.251467 140406029731648 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 00:44:01.920880 140406029731648 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 00:44:02.035013 140406029731648 checkpointing.py:688] Setting up checkpoint logger... I0417 00:44:02.035125 140406029731648 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 00:44:02.035174 140406029731648 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 00:44:02.035378 140406029731648 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 0x7fb23abef650>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 00:44:05.278032 140406029731648 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 00:44:05.278264 140406029731648 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 0x7f9cc477f680>}, handler_registry=None I0417 00:44:05.278501 140406029731648 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9cc477f680>` 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 00:44:05.278550 140406029731648 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9de049d130>` 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 00:44:05.278586 140406029731648 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 0x7f9cc477f680>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9cc477f680>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9de049d130>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9de049d130>}). I0417 00:44:05.278929 140406029731648 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 00:44:05.278998 140406029731648 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7f9de0332020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 00:44:06.505877 140406029731648 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints I0417 00:44:06.925378 140406029731648 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f9de02ccb90> I0417 00:44:06.925548 140406029731648 checkpointing.py:302] Checkpoint manager created! I0417 00:44:07.285155 140406029731648 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 00:44:07.285272 140406029731648 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 00:44:08.697556 140406029731648 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 00:44:08.697668 140406029731648 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 00:44:08.697720 140406029731648 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 00:44:08.697762 140406029731648 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 00:44:08.697798 140406029731648 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 00:44:08.697830 140406029731648 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 00:44:08.697882 140406029731648 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:44:08.697923 140406029731648 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 00:44:08.697959 140406029731648 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:44:08.697994 140406029731648 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:44:08.698027 140406029731648 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 00:44:08.698058 140406029731648 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 00:44:09.101599 140406029731648 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 00:44:09.101690 140406029731648 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 00:44:09.106602 140406029731648 nnx_decoders.py:465] Unknown Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 00:44:09.106657 140406029731648 nnx_decoders.py:465] Unknown Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 00:44:09.122890 140406029731648 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 00:44:09.122949 140406029731648 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 00:44:09.138573 140406029731648 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 00:44:09.138631 140406029731648 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0417 00:44:09.162436 140406029731648 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 00:44:09.162503 140406029731648 attentions.py:1154] attentions/query Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 00:44:09.178298 140406029731648 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 00:44:09.178358 140406029731648 attentions.py:1155] attentions/key Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 00:44:09.193895 140406029731648 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 00:44:09.193969 140406029731648 attentions.py:1156] attentions/value Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 00:44:09.223268 140406029731648 attentions.py:1197] attentions/out Logical: bfloat16[32,128,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 00:44:09.223339 140406029731648 attentions.py:1197] attentions/out Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0417 00:44:09.249005 140406029731648 linears.py:525] linears/x Logical: bfloat16[32,128,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 00:44:09.249070 140406029731648 linears.py:525] linears/x Physical: bfloat16[32,128,7168]....................................... ('fsdp', None, None). I0417 00:44:12.842511 140406029731648 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 00:44:12.851779 140406029731648 metric_logger.py:289] number parameters: 1.104 billion I0417 00:44:16.915385 140406029731648 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 00:44:16.988823 140406029731648 checkpointing.py:798] Waited 0.07342672348022461 seconds for step 0 to finish before starting checkpointing. I0417 00:44:16.991264 140406029731648 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 00:44:16.993062 140406029731648 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 00:44:16.994454 140406029731648 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0. I0417 00:44:17.683292 140406029731648 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 00:44:17.702897 140406029731648 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 00:44:17.702971 140406029731648 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 00:44:17.921661 140277096367872 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0 I0417 00:44:17.973924 140406029731648 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272407s I0417 00:44:17.974674 140406029731648 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.444 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2833364009857178 s) (per-host) I0417 00:44:17.974737 140406029731648 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283416s (batch_requests_ready=0.003252s, total_serialization_initiated=0.279504s, others=0.000660s) I0417 00:44:17.974852 140406029731648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287492s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.287472) I0417 00:44:17.975682 140406029731648 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.98 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0. I0417 00:44:17.975975 140277104760576 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:04:17.975942 I0417 00:44:17.991268 140406029731648 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 00:44:17.991578 140274915690240 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 00:44:17.991737 140406029731648 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776386656.9912455, 'wait_for_prev_duration_secs': 6.318092346191406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776386656.9931014, 'checkpointer_blocking_duration_secs': 0.9829773902893066, 'get_old_steps_start_time': 1776386657.9761074, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776386656.98926, 'checkpoint_manager_blocking_duration_secs': 1.002436876296997} I0417 00:44:17.991858 140406029731648 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 00:44:17.991926 140406029731648 max_utils.py:750] Memstats: After params initialized: I0417 00:44:17.991976 140406029731648 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_0(process=0,(0,0,0,0)) I0417 00:44:17.992007 140406029731648 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_1(process=0,(1,0,0,0)) I0417 00:44:17.992032 140406029731648 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_4(process=0,(0,1,0,0)) I0417 00:44:17.992054 140406029731648 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_5(process=0,(1,1,0,0)) I0417 00:44:18.338177 140406029731648 metric_logger.py:185] completed step: 0, seconds: 3.869, TFLOP/s/device: 0.207, Tokens/s/device: 33.081, total_weights: 4096, loss: 10.874 I0417 00:44:18.341230 140406029731648 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/tensorboard/' I0417 00:44:19.171914 140406029731648 metric_logger.py:185] completed step: 1, seconds: 1.419, TFLOP/s/device: 0.564, Tokens/s/device: 90.179, total_weights: 4096, loss: 10.892 I0417 00:44:19.235909 140406029731648 metric_logger.py:185] completed step: 2, seconds: 0.834, TFLOP/s/device: 0.960, Tokens/s/device: 153.406, total_weights: 4096, loss: 9.969 I0417 00:44:19.306694 140406029731648 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 65.431, Tokens/s/device: 10457.516, total_weights: 4096, loss: 9.125 I0417 00:44:20.099039 140277079582464 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/items I0417 00:44:20.269765 140273858733824 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386659840442858, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776386659840442858, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 00:44:22.027048 2819 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 00:44:24.674498 140275481437952 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0417 00:44:29.048859 140406029731648 metric_logger.py:185] completed step: 4, seconds: 0.065, TFLOP/s/device: 12.339, Tokens/s/device: 1972.083, total_weights: 4096, loss: 8.469 I0417 00:44:29.059148 140406029731648 metric_logger.py:185] completed step: 5, seconds: 0.070, TFLOP/s/device: 11.412, Tokens/s/device: 1823.855, total_weights: 4096, loss: 7.908 I0417 00:44:29.126764 140406029731648 metric_logger.py:185] completed step: 6, seconds: 9.741, TFLOP/s/device: 0.082, Tokens/s/device: 13.140, total_weights: 4096, loss: 7.554 I0417 00:44:29.197324 140406029731648 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 100.917, Tokens/s/device: 16129.032, total_weights: 4096, loss: 7.313 I0417 00:44:29.267789 140406029731648 metric_logger.py:185] completed step: 8, seconds: 0.070, TFLOP/s/device: 11.436, Tokens/s/device: 1827.710, total_weights: 4096, loss: 7.091 I0417 00:44:29.339294 140406029731648 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 00:44:29.339977 140406029731648 checkpointing.py:798] Waited 0.0006895065307617188 seconds for step 9 to finish before starting checkpointing. I0417 00:44:29.342111 140406029731648 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 00:44:54.480597 140274964895488 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.454606s (commit=31.236319s, array_metadata_write=1.218287s) I0417 00:44:54.481660 140277104760576 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 42.935 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.79028940200806 s) (per-host) I0417 00:44:54.481789 140277104760576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.505188s. I0417 00:44:55.201590 140277104760576 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386659840442858, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 00:44:55.965768 140273858733824 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386659840442858, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0417 00:44:56.344131 140277104760576 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000483s. I0417 00:44:57.005003 140277104760576 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 00:44:57.950102 140277104760576 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.622955s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/items I0417 00:44:57.952041 140277104760576 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0/items I0417 00:44:58.502062 140277104760576 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0 I0417 00:44:59.896688 140277104760576 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0`. I0417 00:44:59.897422 140277104760576 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 42.90 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/0 I0417 00:44:59.898794 140277104760576 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 41.922191s. I0417 00:44:59.898989 140274915690240 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 00:44:59.899112 140274915690240 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 00:44:59.899183 140274915690240 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 00:44:59.901017 140274915690240 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 00:44:59.901213 140406029731648 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 00:44:59.901312 140406029731648 checkpoint_manager.py:1452] Waiting for previous save to complete took 30.559204 seconds. If this number is high, consider checkpointing less frequently. I0417 00:44:59.903302 140406029731648 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 00:44:59.905438 140406029731648 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9. I0417 00:45:00.613119 140406029731648 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 00:45:00.613210 140406029731648 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 00:45:00.647423 140406029731648 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035736s I0417 00:45:00.648185 140406029731648 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 34.584 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.044603824615478516 s) (per-host) I0417 00:45:00.648262 140406029731648 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044699s (batch_requests_ready=0.002844s, total_serialization_initiated=0.041141s, others=0.000714s) I0417 00:45:00.648343 140406029731648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.048905s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.048887) I0417 00:45:00.649062 140406029731648 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9. I0417 00:45:00.649393 140274915690240 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:05:00.649357 I0417 00:45:00.654653 140406029731648 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 00:45:00.654957 140276567889664 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 00:45:00.655117 140406029731648 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776386669.3420804, 'wait_for_prev_duration_secs': 30.5592041015625, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776386699.9033415, 'checkpointer_blocking_duration_secs': 0.7461857795715332, 'get_old_steps_start_time': 1776386700.6495504, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776386669.340203, 'checkpoint_manager_blocking_duration_secs': 31.31488037109375} I0417 00:45:00.655228 140406029731648 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 00:45:00.655274 140406029731648 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 00:45:00.841091 140277104760576 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9 I0417 00:45:03.864967 140272235480832 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9/items I0417 00:45:07.496244 140276584675072 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0417 00:45:45.126862 140276576282368 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.130530s (commit=38.941877s, array_metadata_write=1.188653s) I0417 00:45:45.128067 140274915690240 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.477 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.52443885803223 s) (per-host) I0417 00:45:45.128309 140274915690240 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.478806s. I0417 00:45:46.991457 140274915690240 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000315s. I0417 00:45:47.633508 140274915690240 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 00:45:48.567908 140274915690240 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.180349s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9/items I0417 00:45:48.569457 140274915690240 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9/items I0417 00:45:49.101220 140274915690240 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9 I0417 00:45:50.505194 140274915690240 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9`. I0417 00:45:50.506070 140274915690240 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.60 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_03_dropout/checkpoints/9 I0417 00:45:50.507567 140274915690240 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.858061s. I0417 00:45:50.507776 140276567889664 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 00:45:50.507937 140276567889664 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 00:45:50.507998 140276567889664 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 00:45:50.510447 140276567889664 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 00:45:50.510615 140406029731648 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 00:45:50.510786 140406029731648 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 00:45:50.511507 140406029731648 metric_logger.py:185] completed step: 9, seconds: 0.069, TFLOP/s/device: 11.624, Tokens/s/device: 1857.873, 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 00:46:01 UTC 2026 EXIT_CODE=0