XPK Start: Mon Apr 20 17:39:12 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 17:39:36.382261: 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) I0420 17:39:36.560531 132890750306112 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 17:39:45.603024 132890750306112 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 17:39:45.605315 132890750306112 distributed.py:172] Connecting to JAX distributed service on mt-03-dropout-ajaaf-slice-job-0-0.mt-03-dropout-ajaaf:8482 I0420 17:39:46.528267 132890750306112 max_utils.py:284] Jax distributed system initialized! I0420 17:39:52.595985 132890750306112 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 17:39:52.596105 132890750306112 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 17:39:52.596148 132890750306112 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 I0420 17:39:52.596184 132890750306112 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 17:39:53.282899 132890750306112 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 17:39:53.396694 132890750306112 checkpointing.py:688] Setting up checkpoint logger... I0420 17:39:53.396804 132890750306112 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 17:39:53.396858 132890750306112 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 17:39:53.397130 132890750306112 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 0x78dc5264f1a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 17:39:56.342166 132890750306112 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 17:39:56.342394 132890750306112 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 0x78c87c430f20>}, handler_registry=None I0420 17:39:56.342651 132890750306112 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78c87c430f20>` 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`. I0420 17:39:56.342702 132890750306112 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78c87c490b30>` 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`. I0420 17:39:56.342739 132890750306112 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 0x78c87c430f20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78c87c430f20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78c87c490b30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78c87c490b30>}). I0420 17:39:56.343043 132890750306112 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 17:39:56.343125 132890750306112 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78c7bc7ccfe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 17:39:57.324626 132890750306112 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints I0420 17:39:57.789100 132890750306112 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78c87c466db0> I0420 17:39:57.789269 132890750306112 checkpointing.py:302] Checkpoint manager created! I0420 17:39:58.148437 132890750306112 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 17:39:58.148550 132890750306112 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 17:39:59.569622 132890750306112 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 17:39:59.569734 132890750306112 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:39:59.569785 132890750306112 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:39:59.569827 132890750306112 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 17:39:59.569883 132890750306112 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:39:59.569917 132890750306112 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:39:59.569953 132890750306112 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:39:59.569988 132890750306112 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 17:39:59.570021 132890750306112 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:39:59.570053 132890750306112 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:39:59.570097 132890750306112 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 17:39:59.570132 132890750306112 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 17:39:59.983621 132890750306112 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:39:59.983710 132890750306112 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0420 17:39:59.989705 132890750306112 nnx_decoders.py:465] Unknown Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:39:59.989762 132890750306112 nnx_decoders.py:465] Unknown Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0420 17:40:00.006151 132890750306112 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:40:00.006212 132890750306112 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0420 17:40:00.021855 132890750306112 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:40:00.021914 132890750306112 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0420 17:40:00.045792 132890750306112 attentions.py:1154] attentions/query Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:40:00.045859 132890750306112 attentions.py:1154] attentions/query Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0420 17:40:00.061644 132890750306112 attentions.py:1155] attentions/key Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:40:00.061704 132890750306112 attentions.py:1155] attentions/key Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0420 17:40:00.077356 132890750306112 attentions.py:1156] attentions/value Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:40:00.077415 132890750306112 attentions.py:1156] attentions/value Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0420 17:40:00.108029 132890750306112 attentions.py:1197] attentions/out Logical: bfloat16[32,128,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 17:40:00.108115 132890750306112 attentions.py:1197] attentions/out Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0420 17:40:00.134018 132890750306112 linears.py:525] linears/x Logical: bfloat16[32,128,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 17:40:00.134093 132890750306112 linears.py:525] linears/x Physical: bfloat16[32,128,7168]....................................... ('fsdp', None, None). I0420 17:40:03.843477 132890750306112 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. I0420 17:40:03.852779 132890750306112 metric_logger.py:301] number parameters: 1.104 billion I0420 17:40:08.101700 132890750306112 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 17:40:08.174535 132890750306112 checkpointing.py:798] Waited 0.07282423973083496 seconds for step 0 to finish before starting checkpointing. I0420 17:40:08.176721 132890750306112 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:40:08.178784 132890750306112 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 17:40:08.180235 132890750306112 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0. I0420 17:40:08.459508 132890750306112 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 17:40:08.478643 132890750306112 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 17:40:08.478718 132890750306112 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 I0420 17:40:08.726788 132765106280192 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0 I0420 17:40:08.751901 132890750306112 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274772s I0420 17:40:08.752736 132890750306112 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.402 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2855653762817383 s) (per-host) I0420 17:40:08.752809 132890750306112 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285659s (batch_requests_ready=0.002996s, total_serialization_initiated=0.281926s, others=0.000738s) I0420 17:40:08.752897 132890750306112 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289647s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.289629) I0420 17:40:08.753740 132890750306112 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.57 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0. I0420 17:40:08.754027 132765114672896 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:00:08.753993 I0420 17:40:08.766211 132890750306112 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:40:08.766507 132761369175808 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:40:08.766671 132890750306112 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776706808.1767035, 'wait_for_prev_duration_secs': 6.246566772460938e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776706808.1788278, 'checkpointer_blocking_duration_secs': 0.575308084487915, 'get_old_steps_start_time': 1776706808.754161, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776706808.1749227, 'checkpoint_manager_blocking_duration_secs': 0.5917093753814697} I0420 17:40:08.766776 132890750306112 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 17:40:08.766864 132890750306112 max_utils.py:750] Memstats: After params initialized: I0420 17:40:08.766930 132890750306112 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_0(process=0,(0,0,0,0)) I0420 17:40:08.766980 132890750306112 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_1(process=0,(1,0,0,0)) I0420 17:40:08.767022 132890750306112 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_4(process=0,(0,1,0,0)) I0420 17:40:08.767059 132890750306112 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_5(process=0,(1,1,0,0)) I0420 17:40:09.109490 132890750306112 metric_logger.py:196] completed step: 0, seconds: 4.049, TFLOP/s/device: 0.198, Tokens/s/device: 31.611, total_weights: 4096, loss: 10.874, lm_loss: 10.874, perplexity: 52805.141 I0420 17:40:09.113196 132890750306112 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/tensorboard/' I0420 17:40:09.947503 132890750306112 metric_logger.py:196] completed step: 1, seconds: 1.004, TFLOP/s/device: 0.798, Tokens/s/device: 127.479, total_weights: 4096, loss: 10.892, lm_loss: 10.892, perplexity: 53764.488 I0420 17:40:10.012326 132890750306112 metric_logger.py:196] completed step: 2, seconds: 0.839, TFLOP/s/device: 0.955, Tokens/s/device: 152.567, total_weights: 4096, loss: 9.969, lm_loss: 9.969, perplexity: 21350.100 I0420 17:40:10.083040 132890750306112 metric_logger.py:196] completed step: 3, seconds: 0.013, TFLOP/s/device: 63.597, Tokens/s/device: 10164.377, total_weights: 4096, loss: 9.125, lm_loss: 9.125, perplexity: 9186.262 I0420 17:40:10.694098 132764611372800 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/items I0420 17:40:10.853608 132761360783104 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776706810444260189, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776706810444260189, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0420 17:40:11.841254 2780 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 17:40:13.629131 132762442917632 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0420 17:40:19.936335 132890750306112 metric_logger.py:196] completed step: 4, seconds: 0.066, TFLOP/s/device: 12.181, Tokens/s/device: 1946.887, total_weights: 4096, loss: 8.469, lm_loss: 8.469, perplexity: 4765.735 I0420 17:40:19.947541 132890750306112 metric_logger.py:196] completed step: 5, seconds: 0.070, TFLOP/s/device: 11.380, Tokens/s/device: 1818.776, total_weights: 4096, loss: 7.908, lm_loss: 7.908, perplexity: 2719.938 I0420 17:40:20.014107 132890750306112 metric_logger.py:196] completed step: 6, seconds: 9.853, TFLOP/s/device: 0.081, Tokens/s/device: 12.991, total_weights: 4096, loss: 7.554, lm_loss: 7.554, perplexity: 1909.063 I0420 17:40:20.084562 132890750306112 metric_logger.py:196] completed step: 7, seconds: 0.008, TFLOP/s/device: 103.086, Tokens/s/device: 16475.737, total_weights: 4096, loss: 7.313, lm_loss: 7.313, perplexity: 1500.206 I0420 17:40:20.155435 132890750306112 metric_logger.py:196] completed step: 8, seconds: 0.069, TFLOP/s/device: 11.631, Tokens/s/device: 1858.844, total_weights: 4096, loss: 7.091, lm_loss: 7.091, perplexity: 1200.912 I0420 17:40:20.226714 132890750306112 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 17:40:20.227391 132890750306112 checkpointing.py:798] Waited 0.0006861686706542969 seconds for step 9 to finish before starting checkpointing. I0420 17:40:20.229283 132890750306112 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:40:43.102034 132762434524928 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.268355s (commit=30.077309s, array_metadata_write=1.191046s) I0420 17:40:43.103050 132765114672896 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 45.606 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 34.63585305213928 s) (per-host) I0420 17:40:43.103191 132765114672896 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.348542s. I0420 17:40:43.894309 132765114672896 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776706810444260189, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0420 17:40:44.606147 132761360783104 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776706810444260189, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0420 17:40:44.652791 132765114672896 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000424s. I0420 17:40:44.868904 132765114672896 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 17:40:45.492128 132765114672896 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.473543s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/items I0420 17:40:45.493824 132765114672896 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0/items I0420 17:40:46.025304 132765114672896 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0 I0420 17:40:47.392674 132765114672896 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0`. I0420 17:40:47.393501 132765114672896 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 39.21 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/0 I0420 17:40:47.394959 132765114672896 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 38.640309s. I0420 17:40:47.395110 132761369175808 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:40:47.395174 132761369175808 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:40:47.395239 132761369175808 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:40:47.396925 132761369175808 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 17:40:47.397119 132890750306112 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. W0420 17:40:47.397255 132890750306112 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.167969 seconds. If this number is high, consider checkpointing less frequently. I0420 17:40:47.399102 132890750306112 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0420 17:40:47.401273 132890750306112 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9. I0420 17:40:48.144375 132890750306112 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 17:40:48.144466 132890750306112 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 I0420 17:40:48.178482 132890750306112 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035682s I0420 17:40:48.179330 132890750306112 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 34.867 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04424142837524414 s) (per-host) I0420 17:40:48.179390 132890750306112 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044317s (batch_requests_ready=0.002592s, total_serialization_initiated=0.040967s, others=0.000758s) I0420 17:40:48.179466 132890750306112 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.048319s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.048302) I0420 17:40:48.180223 132890750306112 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9. I0420 17:40:48.180567 132761369175808 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:00:48.180529 I0420 17:40:48.184868 132890750306112 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:40:48.185121 132763512411904 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:40:48.185286 132890750306112 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776706820.2292526, 'wait_for_prev_duration_secs': 27.167969465255737, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776706847.3991406, 'checkpointer_blocking_duration_secs': 0.7815704345703125, 'get_old_steps_start_time': 1776706848.180734, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776706820.2276123, 'checkpoint_manager_blocking_duration_secs': 27.9576416015625} I0420 17:40:48.185400 132890750306112 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 17:40:48.185449 132890750306112 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:40:48.392606 132765114672896 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9 I0420 17:40:50.215376 132764594587392 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9/items I0420 17:40:54.367225 132764049323776 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0420 17:41:31.018445 132764032538368 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.242812s (commit=37.804243s, array_metadata_write=1.438569s) I0420 17:41:31.019593 132761369175808 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 36.834 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.88446569442749 s) (per-host) I0420 17:41:31.019721 132761369175808 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.839034s. I0420 17:41:33.794008 132761369175808 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 17:41:33.881442 132761369175808 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000435s. I0420 17:41:34.440527 132761369175808 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.478283s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9/items I0420 17:41:34.442229 132761369175808 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9/items I0420 17:41:35.047578 132761369175808 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9 I0420 17:41:36.494586 132761369175808 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9`. I0420 17:41:36.495442 132761369175808 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.10 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_03_dropout/checkpoints/9 I0420 17:41:36.496753 132761369175808 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.316065s. I0420 17:41:36.496982 132763512411904 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:41:36.497120 132763512411904 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:41:36.497183 132763512411904 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:41:36.498878 132763512411904 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 17:41:36.499036 132890750306112 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. I0420 17:41:36.499194 132890750306112 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:41:36.500398 132890750306112 metric_logger.py:196] completed step: 9, seconds: 0.070, TFLOP/s/device: 11.500, Tokens/s/device: 1837.971, total_weights: 4096, loss: 7.016, lm_loss: 7.016, perplexity: 1113.843 Per train step: Total TFLOPs: 0.80 split as 99.60% learnable weight flops and 0.40% attention flops XPK End: Mon Apr 20 17:41:47 UTC 2026 EXIT_CODE=0