XPK Start: Sat Apr 18 06:12:02 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-18 06:12:41.362871: 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) I0418 06:12:41.542070 134941086005056 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 06:12:50.582936 134941086005056 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 06:12:50.585208 134941086005056 distributed.py:172] Connecting to JAX distributed service on mt-03-dropout-9bcm5-slice-job-0-0.mt-03-dropout-9bcm5:8482 I0418 06:12:52.857116 134941086005056 max_utils.py:284] Jax distributed system initialized! I0418 06:12:58.921642 134941086005056 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 06:12:58.921743 134941086005056 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 06:12:58.921781 134941086005056 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 I0418 06:12:58.921814 134941086005056 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 06:12:59.959110 134941086005056 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 06:12:59.959681 134941086005056 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 06:12:59.959867 134941086005056 checkpointing.py:688] Setting up checkpoint logger... I0418 06:12:59.959918 134941086005056 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 06:12:59.959960 134941086005056 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 06:12:59.960292 134941086005056 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 0x7ab9b3f38800>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 06:13:03.224701 134941086005056 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0418 06:13:03.224990 134941086005056 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 0x7ab8a7fa9e20>}, handler_registry=None I0418 06:13:03.225236 134941086005056 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab8a7fa9e20>` 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`. I0418 06:13:03.225285 134941086005056 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab8a7d4ef30>` 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`. I0418 06:13:03.225320 134941086005056 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 0x7ab8a7fa9e20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab8a7fa9e20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab8a7d4ef30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab8a7d4ef30>}). I0418 06:13:03.225639 134941086005056 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 06:13:03.225708 134941086005056 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7aa5443df740> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 06:13:06.325318 134941086005056 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints I0418 06:13:07.187954 134941086005056 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/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ab8a7d4cb00> I0418 06:13:07.188130 134941086005056 checkpointing.py:302] Checkpoint manager created! I0418 06:13:07.519793 134941086005056 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 06:13:07.519913 134941086005056 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0418 06:13:07.903151 134941086005056 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 06:13:07.903241 134941086005056 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0418 06:13:07.919199 134941086005056 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,128,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 06:13:07.919256 134941086005056 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,128,2048]....................................... ('fsdp', None, None). I0418 06:13:07.942542 134941086005056 attentions.py:1154] attentions/query Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 06:13:07.942611 134941086005056 attentions.py:1154] attentions/query Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0418 06:13:07.958662 134941086005056 attentions.py:1155] attentions/key Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 06:13:07.958723 134941086005056 attentions.py:1155] attentions/key Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0418 06:13:07.974692 134941086005056 attentions.py:1156] attentions/value Logical: bfloat16[32,128,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 06:13:07.974748 134941086005056 attentions.py:1156] attentions/value Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0418 06:13:08.000047 134941086005056 attentions.py:1197] attentions/out Logical: bfloat16[32,128,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 06:13:08.000114 134941086005056 attentions.py:1197] attentions/out Physical: bfloat16[32,128,16,128]..................................... ('fsdp', None, None, None). I0418 06:13:08.025915 134941086005056 linears.py:525] linears/x Logical: bfloat16[32,128,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 06:13:08.025984 134941086005056 linears.py:525] linears/x Physical: bfloat16[32,128,7168]....................................... ('fsdp', None, None). I0418 06:13:08.248549 134941086005056 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0418 06:13:08.248657 134941086005056 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 06:13:09.672031 134941086005056 maxtext_utils.py:1821] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 06:13:09.672153 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 06:13:09.672207 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 06:13:09.672265 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0418 06:13:09.672332 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 06:13:09.672398 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 06:13:09.672460 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0418 06:13:09.672513 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0418 06:13:09.672554 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0418 06:13:09.672590 134941086005056 maxtext_utils.py:1821] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0418 06:13:09.672637 134941086005056 maxtext_utils.py:1821] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 06:13:09.672685 134941086005056 maxtext_utils.py:1821] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 06:13:10.190197 134941086005056 train.py:157] train/xent Logical: float32[32,128]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0418 06:13:10.190312 134941086005056 train.py:157] train/xent Physical: float32[32,128]............................................. ('fsdp', None). I0418 06:13:10.205679 134941086005056 train.py:164] train/z_loss Logical: float32[32,128]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0418 06:13:10.205739 134941086005056 train.py:164] train/z_loss Physical: float32[32,128]............................................. ('fsdp', None). I0418 06:13:13.902062 134941086005056 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. I0418 06:13:13.910351 134941086005056 metric_logger.py:301] number parameters: 1.104 billion I0418 06:13:18.228398 134941086005056 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0418 06:13:18.302352 134941086005056 checkpointing.py:798] Waited 0.07394099235534668 seconds for step 0 to finish before starting checkpointing. I0418 06:13:18.304928 134941086005056 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 06:13:18.306895 134941086005056 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0418 06:13:18.308311 134941086005056 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0. I0418 06:13:19.011827 134941086005056 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 06:13:19.025618 134941086005056 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 06:13:19.025686 134941086005056 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 I0418 06:13:19.264041 134812463503104 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0 I0418 06:13:19.294610 134941086005056 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269758s I0418 06:13:19.295471 134941086005056 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.591 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2759103775024414 s) (per-host) I0418 06:13:19.295550 134941086005056 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.276010s (batch_requests_ready=0.002030s, total_serialization_initiated=0.273200s, others=0.000780s) I0418 06:13:19.295635 134941086005056 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.280018s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.279999) I0418 06:13:19.296347 134941086005056 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.99 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0. I0418 06:13:19.296670 134812471895808 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 06:33:19.296633 I0418 06:13:19.312966 134941086005056 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 06:13:19.313261 134810307692288 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0418 06:13:19.313413 134941086005056 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776492798.3049095, 'wait_for_prev_duration_secs': 6.175041198730469e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776492798.3069327, 'checkpointer_blocking_duration_secs': 0.98988938331604, 'get_old_steps_start_time': 1776492799.296859, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776492798.3028276, 'checkpoint_manager_blocking_duration_secs': 1.0105478763580322} I0418 06:13:19.313515 134941086005056 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0418 06:13:19.313564 134941086005056 max_utils.py:750] Memstats: After params initialized: I0418 06:13:19.313613 134941086005056 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_0(process=0,(0,0,0,0)) I0418 06:13:19.313647 134941086005056 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_1(process=0,(1,0,0,0)) I0418 06:13:19.313671 134941086005056 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_4(process=0,(0,1,0,0)) I0418 06:13:19.313694 134941086005056 max_utils.py:756] Using (GB) 0.41 / 31.25 (1.312000%) on TPU_5(process=0,(1,1,0,0)) I0418 06:13:19.647353 134941086005056 metric_logger.py:196] completed step: 0, seconds: 4.135, TFLOP/s/device: 0.194, Tokens/s/device: 30.957, total_weights: 4096, loss: 10.889, lm_loss: 10.889, perplexity: 53560.699 I0418 06:13:19.650592 134941086005056 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/tensorboard/' I0418 06:13:20.038599 134941086005056 metric_logger.py:196] completed step: 1, seconds: 1.417, TFLOP/s/device: 0.565, Tokens/s/device: 90.322, total_weights: 4096, loss: 10.902, lm_loss: 10.902, perplexity: 54298.523 I0418 06:13:20.107241 134941086005056 metric_logger.py:196] completed step: 2, seconds: 0.392, TFLOP/s/device: 2.044, Tokens/s/device: 326.683, total_weights: 4096, loss: 9.956, lm_loss: 9.956, perplexity: 21082.051 I0418 06:13:20.178215 134941086005056 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 85.968, Tokens/s/device: 13739.802, total_weights: 4096, loss: 9.139, lm_loss: 9.139, perplexity: 9314.691 I0418 06:13:22.736221 134811968595712 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/items I0418 06:13:23.331653 134808755762944 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776492802481509550, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776492802481509550, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0418 06:13:24.286565 2840 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 06:13:27.764298 134810324477696 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0418 06:13:31.391950 134941086005056 metric_logger.py:196] completed step: 4, seconds: 0.069, TFLOP/s/device: 11.526, Tokens/s/device: 1842.124, total_weights: 4096, loss: 8.459, lm_loss: 8.459, perplexity: 4715.493 I0418 06:13:31.400627 134941086005056 metric_logger.py:196] completed step: 5, seconds: 0.071, TFLOP/s/device: 11.264, Tokens/s/device: 1800.307, total_weights: 4096, loss: 7.924, lm_loss: 7.924, perplexity: 2762.373 I0418 06:13:31.469730 134941086005056 metric_logger.py:196] completed step: 6, seconds: 11.212, TFLOP/s/device: 0.071, Tokens/s/device: 11.416, total_weights: 4096, loss: 7.553, lm_loss: 7.553, perplexity: 1906.617 I0418 06:13:31.540468 134941086005056 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 119.641, Tokens/s/device: 19121.601, total_weights: 4096, loss: 7.267, lm_loss: 7.267, perplexity: 1431.653 I0418 06:13:31.611136 134941086005056 metric_logger.py:196] completed step: 8, seconds: 0.070, TFLOP/s/device: 11.449, Tokens/s/device: 1829.852, total_weights: 4096, loss: 7.111, lm_loss: 7.111, perplexity: 1225.983 I0418 06:13:31.681064 134941086005056 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0418 06:13:31.681732 134941086005056 checkpointing.py:798] Waited 0.0006837844848632812 seconds for step 9 to finish before starting checkpointing. I0418 06:13:31.683932 134941086005056 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 06:13:57.710382 134810316084992 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.432522s (commit=31.445370s, array_metadata_write=1.987152s) I0418 06:13:57.711493 134812471895808 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 40.825 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.69190692901611 s) (per-host) I0418 06:13:57.711620 134812471895808 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.414836s. I0418 06:13:58.831760 134812471895808 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776492802481509550, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0418 06:13:59.989761 134808755762944 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776492802481509550, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0418 06:14:00.201499 134812471895808 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0418 06:14:00.366747 134812471895808 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000250s. I0418 06:14:01.148766 134812471895808 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.191621s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/items I0418 06:14:01.150597 134812471895808 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0/items I0418 06:14:01.724365 134812471895808 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0 I0418 06:14:04.465905 134812471895808 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0`. I0418 06:14:04.466723 134812471895808 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.16 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/0 I0418 06:14:04.468108 134812471895808 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.171322s. I0418 06:14:04.468314 134810307692288 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 06:14:04.468429 134810307692288 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0418 06:14:04.468494 134810307692288 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 06:14:04.470867 134810307692288 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 06:14:04.471071 134941086005056 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. W0418 06:14:04.471212 134941086005056 checkpoint_manager.py:1452] Waiting for previous save to complete took 32.787285 seconds. If this number is high, consider checkpointing less frequently. I0418 06:14:04.473042 134941086005056 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0418 06:14:04.475036 134941086005056 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9. I0418 06:14:04.778112 134941086005056 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 06:14:04.778206 134941086005056 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 I0418 06:14:04.811954 134941086005056 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034586s I0418 06:14:04.812804 134941086005056 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 39.114 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03943824768066406 s) (per-host) I0418 06:14:04.812897 134941086005056 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039550s (batch_requests_ready=0.001781s, total_serialization_initiated=0.036982s, others=0.000787s) I0418 06:14:04.812979 134941086005056 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043854s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.043838) I0418 06:14:04.813685 134941086005056 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9. I0418 06:14:04.813994 134810307692288 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 06:34:04.813947 I0418 06:14:04.817629 134941086005056 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 06:14:04.817929 134811406546688 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0418 06:14:04.818090 134941086005056 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776492811.6838937, 'wait_for_prev_duration_secs': 32.78728532791138, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776492844.4730804, 'checkpointer_blocking_duration_secs': 0.34105443954467773, 'get_old_steps_start_time': 1776492844.814157, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776492811.6820738, 'checkpoint_manager_blocking_duration_secs': 33.13598132133484} I0418 06:14:04.818201 134941086005056 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0418 06:14:04.818245 134941086005056 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 06:14:05.017867 134812471895808 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9 I0418 06:14:08.101917 134810324477696 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9/items I0418 06:14:12.738090 134811434186496 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0418 06:14:51.317269 134811414939392 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.653725s (commit=39.364986s, array_metadata_write=2.288739s) I0418 06:14:51.318322 134810307692288 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.937 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.54490804672241 s) (per-host) I0418 06:14:51.318448 134810307692288 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.504343s. I0418 06:14:54.097775 134810307692288 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000216s. I0418 06:14:54.333076 134810307692288 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0418 06:14:55.294988 134810307692288 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.206927s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9/items I0418 06:14:55.296813 134810307692288 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9/items I0418 06:14:55.801195 134810307692288 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9 I0418 06:14:57.619624 134810307692288 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9`. I0418 06:14:57.620447 134810307692288 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 53.15 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141/linen_xpk_feat_nnx_trainstate_and_training_loop_20260418_060141_03_dropout/checkpoints/9 I0418 06:14:57.621905 134810307692288 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.807798s. I0418 06:14:57.622092 134811406546688 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 06:14:57.622212 134811406546688 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0418 06:14:57.622274 134811406546688 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 06:14:57.624380 134811406546688 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 06:14:57.624551 134941086005056 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. I0418 06:14:57.624698 134941086005056 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 06:14:57.625681 134941086005056 metric_logger.py:196] completed step: 9, seconds: 0.071, TFLOP/s/device: 11.334, Tokens/s/device: 1811.440, total_weights: 4096, loss: 6.981, lm_loss: 6.981, perplexity: 1075.867 Per train step: Total TFLOPs: 0.80 split as 99.60% learnable weight flops and 0.40% attention flops XPK End: Sat Apr 18 06:15:08 UTC 2026 EXIT_CODE=0