XPK Start: Wed Apr 22 17:27:07 UTC 2026 2026-04-22 17:27:11.735295: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776878831.748195 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776878831.752165 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776878831.763657 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776878831.763682 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776878831.763685 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776878831.763687 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-22 17:27:31.726737: 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) I0422 17:27:32.255257 132985325864768 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-22 17:27:41,297:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0422 17:27:41.297092 132985325864768 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-22 17:27:41,299:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-false-rs-p6y61-slice-job-0-0.mt-16-zarr3-false-rs-p6y61:8482 I0422 17:27:41.299423 132985325864768 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-false-rs-p6y61-slice-job-0-0.mt-16-zarr3-false-rs-p6y61:8482 I0422 17:27:42.372778 132985325864768 max_utils.py:284] Jax distributed system initialized! I0422 17:27:48.113127 132985325864768 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 17:27:48.113231 132985325864768 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 17:27:48.113272 132985325864768 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0422 17:27:48.113316 132985325864768 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 17:27:48.735778 132985325864768 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 17:27:48.736430 132985325864768 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 17:27:48.736629 132985325864768 checkpointing.py:688] Setting up checkpoint logger... I0422 17:27:48.736685 132985325864768 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0422 17:27:48.736731 132985325864768 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0422 17:27:48.737102 132985325864768 base_pytree_checkpoint_handler.py:415] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x78f29319f200>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 17:27:50.481873 132985325864768 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 17:27:50.482243 132985325864768 checkpoint_manager.py:709] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78e0582edc70>}, handler_registry=None I0422 17:27:50.482481 132985325864768 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78e0582edc70>` 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`. I0422 17:27:50.482535 132985325864768 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78e0582f2cf0>` 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`. I0422 17:27:50.482571 132985325864768 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 0x78e0582edc70>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78e0582edc70>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78e0582f2cf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78e0582f2cf0>}). I0422 17:27:50.483093 132985325864768 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0422 17:27:50.483173 132985325864768 async_checkpointer.py:177] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78debc4f0180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 17:27:51.955977 132866574882560 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776878588884788455, 'commit_timestamp_nsecs': 1776878636681852795, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0422 17:27:52.208879 132866574882560 checkpoint_manager.py:1877] Missing metrics for step 0 I0422 17:27:52.395737 132985325864768 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints I0422 17:27:52.668780 132985325864768 checkpoint_manager.py:929] [process=3][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=5), 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_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78e0592e53a0> I0422 17:27:52.669000 132985325864768 checkpointing.py:302] Checkpoint manager created! I0422 17:27:53.565106 132985325864768 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 17:27:53.565211 132985325864768 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 17:27:53.935623 132985325864768 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:27:53.935720 132985325864768 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 17:27:53.951596 132985325864768 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:27:53.951658 132985325864768 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 17:27:53.974621 132985325864768 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:27:53.974694 132985325864768 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 17:27:53.990834 132985325864768 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:27:53.990897 132985325864768 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 17:27:54.006820 132985325864768 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:27:54.006881 132985325864768 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 17:27:54.031704 132985325864768 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 17:27:54.031783 132985325864768 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 17:27:54.051403 132985325864768 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 17:27:54.051471 132985325864768 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 17:27:54.241487 132985325864768 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 17:27:54.241598 132985325864768 checkpointing.py:582] restoring from this run's directory step 9 I0422 17:27:54.736629 132985325864768 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/9. I0422 17:27:56.201569 1396 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 17:28:02.108647 132985325864768 jax_array_handlers.py:837] [process=3] /jax/orbax/read/worker/io/requested throughput: 267.288 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 5.9122560024261475 s) (per-host) I0422 17:28:02.113576 132985325864768 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/read/gbytes_per_sec: 248.580 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.366540908813477 s) (per-host) I0422 17:28:03.399762 132985325864768 checkpointer.py:318] Finished restoring checkpoint in 8.92 seconds from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/9. I0422 17:28:03.399921 132985325864768 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776878874.4802537, 'checkpointer_duration_secs': 8.919634342193604, 'checkpoint_manager_start_time': 1776878874.2418509, 'checkpoint_manager_duration_secs': 9.158039808273315} fsdp: 32 I0422 17:28:03.805683 132985325864768 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 17:28:03.805815 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 17:28:03.805870 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 17:28:03.805946 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0422 17:28:03.806029 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 17:28:03.806092 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 17:28:03.806170 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 17:28:03.806245 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0422 17:28:03.806302 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0422 17:28:03.806354 132985325864768 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 17:28:03.806425 132985325864768 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0422 17:28:03.806503 132985325864768 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0422 17:28:04.307029 132985325864768 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:28:04.307123 132985325864768 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0422 17:28:04.322147 132985325864768 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:28:04.322206 132985325864768 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0422 17:28:14.940487 132985325864768 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0422 17:28:14.941323 132985325864768 metric_logger.py:289] number parameters: 1.104 billion I0422 17:28:26.121523 132985325864768 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0422 17:28:26.277497 132985325864768 checkpointing.py:798] Waited 0.15595626831054688 seconds for step 10 to finish before starting checkpointing. I0422 17:28:26.279916 132985325864768 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:28:26.281939 132985325864768 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 10 I0422 17:28:26.284605 132985325864768 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/10. I0422 17:28:26.905314 132985325864768 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 17:28:26.906376 132985325864768 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 17:28:26.906435 132985325864768 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 I0422 17:28:27.160092 132985325864768 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.254872s I0422 17:28:27.160266 132985325864768 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.843 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5425059795379639 s) (per-host) I0422 17:28:27.160321 132985325864768 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.542572s (batch_requests_ready=0.285573s, total_serialization_initiated=0.256924s, others=0.000075s) I0422 17:28:27.160412 132985325864768 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.546768s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.546750) I0422 17:28:27.161312 132860628104960 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:28:27.161465 132985325864768 async_checkpointer.py:561] Finished blocking save. Time taken: 0.879455s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/10. I0422 17:28:27.163441 132985325864768 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:28:27.163728 132864989447936 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:28:27.163905 132985325864768 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776878906.2798977, 'wait_for_prev_duration_secs': 7.748603820800781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776878906.2819774, 'checkpointer_blocking_duration_secs': 0.8796873092651367, 'get_old_steps_start_time': 1776878907.1616838, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776878906.2780995, 'checkpoint_manager_blocking_duration_secs': 0.8857686519622803} I0422 17:28:27.164032 132985325864768 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0422 17:28:27.164087 132985325864768 max_utils.py:750] Memstats: After params initialized: I0422 17:28:27.164135 132985325864768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0422 17:28:27.164167 132985325864768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0422 17:28:27.164195 132985325864768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0422 17:28:27.164220 132985325864768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0422 17:28:27.432654 132985325864768 metric_logger.py:185] completed step: 10, seconds: 11.180, TFLOP/s/device: 1.215, Tokens/s/device: 183.183, total_weights: 65536, loss: 6.833 I0422 17:28:27.605898 132985325864768 metric_logger.py:185] completed step: 11, seconds: 1.310, TFLOP/s/device: 10.374, Tokens/s/device: 1563.664, total_weights: 65536, loss: 6.528 I0422 17:28:27.753290 132985325864768 metric_logger.py:185] completed step: 12, seconds: 0.026, TFLOP/s/device: 518.969, Tokens/s/device: 78224.667, total_weights: 65536, loss: 6.265 I0422 17:28:27.900755 132985325864768 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.777, Tokens/s/device: 13381.423, total_weights: 65536, loss: 6.045 I0422 17:28:27.905494 132985325864768 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0422 17:28:28.195420 132985325864768 checkpointing.py:798] Waited 0.2898898124694824 seconds for step 15 to finish before starting checkpointing. I0422 17:28:28.198097 132985325864768 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:28:31.722999 132860619712256 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/10/items/array_metadatas/process_3 I0422 17:29:11.759703 132860628104960 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 34.992 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.141905784606934 s) (per-host) I0422 17:29:11.759833 132860628104960 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.598277s. I0422 17:29:16.547970 132860628104960 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 49.386374s. I0422 17:29:16.548268 132864989447936 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:29:16.548388 132864989447936 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:29:16.548438 132864989447936 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:29:16.550019 132864989447936 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 17:29:16.550209 132985325864768 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0422 17:29:16.550339 132985325864768 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.352245 seconds. If this number is high, consider checkpointing less frequently. I0422 17:29:16.552311 132985325864768 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 15 I0422 17:29:16.556132 132985325864768 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/15. I0422 17:29:17.107270 132985325864768 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 17:29:17.107369 132985325864768 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 I0422 17:29:17.152652 132985325864768 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046568s I0422 17:29:17.152819 132985325864768 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.129 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3007504940032959 s) (per-host) I0422 17:29:17.152870 132985325864768 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.300811s (batch_requests_ready=0.252280s, total_serialization_initiated=0.048463s, others=0.000069s) I0422 17:29:17.152974 132985325864768 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.305717s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.305703) I0422 17:29:17.153822 132860066199296 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:29:17.154001 132985325864768 async_checkpointer.py:561] Finished blocking save. Time taken: 0.601615s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/15. I0422 17:29:17.589899 132985325864768 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:29:17.590278 132864989447936 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:29:17.590458 132985325864768 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776878908.1980693, 'wait_for_prev_duration_secs': 48.35224533081055, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776878956.5523522, 'checkpointer_blocking_duration_secs': 0.601757287979126, 'get_old_steps_start_time': 1776878957.1541276, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776878908.1961453, 'checkpoint_manager_blocking_duration_secs': 49.39427065849304} I0422 17:29:17.590584 132985325864768 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0422 17:29:17.591215 132985325864768 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.227, Tokens/s/device: 13901.427, total_weights: 65536, loss: 5.864 I0422 17:29:17.597375 132985325864768 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 92.078, Tokens/s/device: 13879.006, total_weights: 65536, loss: 5.719 I0422 17:29:35.402987 132985325864768 metric_logger.py:185] completed step: 16, seconds: 49.691, TFLOP/s/device: 0.273, Tokens/s/device: 41.215, total_weights: 65536, loss: 5.605 I0422 17:29:35.550419 132985325864768 metric_logger.py:185] completed step: 17, seconds: 17.659, TFLOP/s/device: 0.769, Tokens/s/device: 115.973, total_weights: 65536, loss: 5.517 I0422 17:29:35.697823 132985325864768 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.319, Tokens/s/device: 13463.144, total_weights: 65536, loss: 5.448 I0422 17:29:35.844970 132985325864768 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0422 17:29:35.845644 132985325864768 checkpointing.py:798] Waited 0.0007278919219970703 seconds for step 19 to finish before starting checkpointing. I0422 17:29:35.847729 132985325864768 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:29:38.800233 132860619712256 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/15/items/array_metadatas/process_3 I0422 17:30:15.262776 132860066199296 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 27.043 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 58.41065835952759 s) (per-host) I0422 17:30:15.262895 132860066199296 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 58.108945s. I0422 17:30:24.117916 132860066199296 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 66.963951s. I0422 17:30:24.118227 132864989447936 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:30:24.118369 132864989447936 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:30:24.118421 132864989447936 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:30:24.120297 132864989447936 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 17:30:24.120492 132985325864768 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0422 17:30:24.120614 132985325864768 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.272896 seconds. If this number is high, consider checkpointing less frequently. I0422 17:30:24.122001 132985325864768 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 19 I0422 17:30:24.125523 132985325864768 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/19. I0422 17:30:24.794803 132985325864768 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 17:30:24.794901 132985325864768 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 I0422 17:30:24.840341 132985325864768 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046580s I0422 17:30:24.840509 132985325864768 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.079 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30368924140930176 s) (per-host) I0422 17:30:24.840558 132985325864768 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.303748s (batch_requests_ready=0.255008s, total_serialization_initiated=0.048676s, others=0.000064s) I0422 17:30:24.840649 132985325864768 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.307972s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=0.307959) I0422 17:30:24.841546 132865484388096 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:30:24.841704 132985325864768 async_checkpointer.py:561] Finished blocking save. Time taken: 0.719637s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/19. I0422 17:30:25.270763 132985325864768 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:30:25.271143 132864989447936 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:30:25.271288 132985325864768 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776878975.8477013, 'wait_for_prev_duration_secs': 48.27289605140686, 'time_between_consecutive_saves_sec': 19.297640800476074, 'checkpointer_blocking_start_time': 1776879024.1220396, 'checkpointer_blocking_duration_secs': 0.719773530960083, 'get_old_steps_start_time': 1776879024.8418267, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776878975.8458736, 'checkpoint_manager_blocking_duration_secs': 49.4253830909729} I0422 17:30:25.271452 132985325864768 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0422 17:30:25.271499 132985325864768 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:30:29.658434 132865970902784 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_16_zarr3_false/checkpoints/19/items/array_metadatas/process_3 I0422 17:31:05.480435 132865484388096 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 38.580 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.943583726882935 s) (per-host) I0422 17:31:05.480543 132865484388096 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.638882s. I0422 17:31:14.269153 132865484388096 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 49.427475s. I0422 17:31:14.269425 132864989447936 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:31:14.269541 132864989447936 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:31:14.269583 132864989447936 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:31:14.271110 132864989447936 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 17:31:14.271275 132985325864768 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0422 17:31:14.271422 132985325864768 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:31:14.272076 132985325864768 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.137, Tokens/s/device: 13887.947, total_weights: 65536, loss: 5.391 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Wed Apr 22 17:31:28 UTC 2026 EXIT_CODE=0