XPK Start: Sat Apr 18 14:25:56 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 14:26:20.546504: 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 14:26:21.086435 138541847643968 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 14:26:30.128683 138541847643968 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 14:26:30.130939 138541847643968 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-al0oj-slice-job-0-0.mt-16-zarr3-true-sv-al0oj:8482 I0418 14:26:31.259155 138541847643968 max_utils.py:284] Jax distributed system initialized! I0418 14:26:37.540712 138541847643968 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 14:26:37.540812 138541847643968 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 14:26:37.540851 138541847643968 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 14:26:37.540884 138541847643968 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 14:26:38.236305 138541847643968 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 14:26:38.236584 138541847643968 checkpointing.py:677] Setting up checkpoint logger... I0418 14:26:38.236639 138541847643968 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 14:26:38.236683 138541847643968 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 14:26:38.237024 138541847643968 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 0x7e0011f4c0b0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 14:26:41.192266 138541847643968 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 14:26:41.192502 138541847643968 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dff05d8bef0>}, handler_registry=None I0418 14:26:41.192741 138541847643968 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dff05d8bef0>` 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 14:26:41.192789 138541847643968 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7debc04c8590>` 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 14:26:41.192825 138541847643968 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 0x7dff05d8bef0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dff05d8bef0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7debc04c8590>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7debc04c8590>}). I0418 14:26:41.193159 138541847643968 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 14:26:41.193227 138541847643968 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7dead84e7c40> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 14:26:43.626662 138541847643968 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints I0418 14:26:43.628995 138541847643968 checkpoint_manager.py:929] [process=7][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_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e0071bcf680> I0418 14:26:43.629123 138541847643968 checkpointing.py:301] Checkpoint manager created! I0418 14:26:44.563031 138541847643968 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 14:26:44.563174 138541847643968 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:26:44.940828 138541847643968 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:26:44.940917 138541847643968 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:26:44.957093 138541847643968 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:26:44.957156 138541847643968 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:26:44.981297 138541847643968 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 14:26:44.981369 138541847643968 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:26:44.997650 138541847643968 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 14:26:44.997709 138541847643968 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:26:45.013957 138541847643968 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 14:26:45.014017 138541847643968 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:26:45.038759 138541847643968 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 14:26:45.038826 138541847643968 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:26:45.059714 138541847643968 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 14:26:45.059785 138541847643968 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 14:26:45.274310 138541847643968 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 14:26:45.274413 138541847643968 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 14:26:46.694539 138541847643968 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 14:26:46.694659 138541847643968 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 14:26:46.694710 138541847643968 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 14:26:46.694765 138541847643968 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0418 14:26:46.694887 138541847643968 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 14:26:46.694945 138541847643968 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 14:26:46.695014 138541847643968 maxtext_utils.py:1654] 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 14:26:46.695111 138541847643968 maxtext_utils.py:1654] 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 14:26:46.695158 138541847643968 maxtext_utils.py:1654] 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 14:26:46.695198 138541847643968 maxtext_utils.py:1654] 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 14:26:46.695246 138541847643968 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 14:26:46.695315 138541847643968 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 14:26:47.180743 138541847643968 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:26:47.180837 138541847643968 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:26:47.196404 138541847643968 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:26:47.196467 138541847643968 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:26:58.089543 138541847643968 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. I0418 14:26:58.090439 138541847643968 metric_logger.py:301] number parameters: 1.104 billion I0418 14:27:09.321474 138541847643968 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 14:27:09.788436 138541847643968 checkpointing.py:776] Waited 0.46693944931030273 seconds for step 0 to finish before starting checkpointing. I0418 14:27:09.791256 138541847643968 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:27:09.793090 138541847643968 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0 I0418 14:27:09.794474 138541847643968 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/0. I0418 14:27:10.529581 138541847643968 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 14:27:10.532370 138541847643968 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:27:10.532430 138541847643968 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 14:27:10.810620 138541847643968 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.281117s I0418 14:27:10.810813 138541847643968 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.380 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2867457866668701 s) (per-host) I0418 14:27:10.810886 138541847643968 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286821s (batch_requests_ready=0.002330s, total_serialization_initiated=0.284405s, others=0.000085s) I0418 14:27:10.811006 138541847643968 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291016s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.290997) I0418 14:27:10.813183 138541847643968 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/0. I0418 14:27:10.813480 138413389584128 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:47:10.813444 I0418 14:27:10.815442 138541847643968 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:27:10.815724 138411822278400 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:27:10.815877 138541847643968 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776522429.7912366, 'wait_for_prev_duration_secs': 6.151199340820312e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776522429.7931285, 'checkpointer_blocking_duration_secs': 1.0205039978027344, 'get_old_steps_start_time': 1776522430.813658, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776522429.788931, 'checkpoint_manager_blocking_duration_secs': 1.0269103050231934} I0418 14:27:10.815986 138541847643968 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 14:27:10.816035 138541847643968 max_utils.py:750] Memstats: After params initialized: I0418 14:27:10.816093 138541847643968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0418 14:27:10.816129 138541847643968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0418 14:27:10.816156 138541847643968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0418 14:27:10.816180 138541847643968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0418 14:27:11.136204 138541847643968 metric_logger.py:196] completed step: 0, seconds: 11.231, TFLOP/s/device: 1.210, Tokens/s/device: 182.354, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:27:11.329174 138541847643968 metric_logger.py:196] completed step: 1, seconds: 1.813, TFLOP/s/device: 7.494, Tokens/s/device: 1129.512, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:27:11.752847 138541847643968 metric_logger.py:196] completed step: 2, seconds: 0.046, TFLOP/s/device: 297.103, Tokens/s/device: 44782.647, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719 I0418 14:27:11.900320 138541847643968 metric_logger.py:196] completed step: 3, seconds: 0.424, TFLOP/s/device: 32.050, Tokens/s/device: 4830.963, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506 I0418 14:27:11.904674 138541847643968 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0418 14:27:12.194554 138541847643968 checkpointing.py:776] Waited 0.2898423671722412 seconds for step 5 to finish before starting checkpointing. I0418 14:27:12.197607 138541847643968 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:27:14.160564 2874 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 14:27:16.208822 138411830671104 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/0/items/array_metadatas/process_7 I0418 14:27:48.027951 138413389584128 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 42.118 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.50384259223938 s) (per-host) I0418 14:27:48.028098 138413389584128 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.214508s. I0418 14:27:56.162455 138413389584128 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 45.348850s. I0418 14:27:56.162727 138411822278400 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:27:56.162864 138411822278400 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:27:56.162915 138411822278400 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:27:56.164794 138411822278400 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 14:27:56.164995 138541847643968 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 14:27:56.165089 138541847643968 checkpoint_manager.py:1452] Waiting for previous save to complete took 43.967470 seconds. If this number is high, consider checkpointing less frequently. I0418 14:27:56.167159 138541847643968 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 5 I0418 14:27:56.169795 138541847643968 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/5. I0418 14:27:57.311815 138541847643968 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:27:57.311908 138541847643968 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 14:27:57.348499 138541847643968 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037698s I0418 14:27:57.348667 138541847643968 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 37.382 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04126477241516113 s) (per-host) I0418 14:27:57.348715 138541847643968 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.041323s (batch_requests_ready=0.001944s, total_serialization_initiated=0.039316s, others=0.000063s) I0418 14:27:57.348802 138541847643968 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.046408s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.046392) I0418 14:27:57.349591 138541847643968 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.18 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/5. I0418 14:27:57.349908 138411822278400 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:47:57.349870 I0418 14:27:57.360032 138541847643968 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:27:57.360325 138412904412928 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:27:57.360468 138541847643968 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776522432.1975787, 'wait_for_prev_duration_secs': 43.96747040748596, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776522476.1672003, 'checkpointer_blocking_duration_secs': 1.1828532218933105, 'get_old_steps_start_time': 1776522477.350089, 'get_old_steps_duration_secs': 3.719329833984375e-05, 'checkpoint_manager_blocking_start_time': 1776522432.19522, 'checkpoint_manager_blocking_duration_secs': 45.16520619392395} I0418 14:27:57.360599 138541847643968 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0418 14:27:57.361406 138541847643968 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.850, Tokens/s/device: 13392.448, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355 I0418 14:27:57.367716 138541847643968 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.216, Tokens/s/device: 13899.729, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467 I0418 14:28:02.684328 138411830671104 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/5/items/array_metadatas/process_7 I0418 14:28:20.373466 138541847643968 metric_logger.py:196] completed step: 6, seconds: 45.462, TFLOP/s/device: 0.299, Tokens/s/device: 45.049, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740 I0418 14:28:20.520910 138541847643968 metric_logger.py:196] completed step: 7, seconds: 22.860, TFLOP/s/device: 0.594, Tokens/s/device: 89.590, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151 I0418 14:28:20.668371 138541847643968 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.272, Tokens/s/device: 13456.067, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098 I0418 14:28:20.814986 138541847643968 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 14:28:20.815681 138541847643968 checkpointing.py:776] Waited 0.0007152557373046875 seconds for step 9 to finish before starting checkpointing. I0418 14:28:20.817805 138541847643968 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:28:39.263735 138411822278400 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.649 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.95629358291626 s) (per-host) I0418 14:28:39.263869 138411822278400 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.913838s. I0418 14:28:45.813878 138411822278400 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 48.463836s. I0418 14:28:45.814198 138412904412928 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:28:45.814327 138412904412928 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:28:45.814375 138412904412928 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:28:45.815637 138412904412928 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0418 14:28:45.818290 138541847643968 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0418 14:28:45.818413 138541847643968 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.000613 seconds. If this number is high, consider checkpointing less frequently. I0418 14:28:45.819670 138541847643968 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9 I0418 14:28:45.821803 138541847643968 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/9. I0418 14:28:46.967771 138541847643968 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:28:46.967865 138541847643968 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 14:28:46.993689 138541847643968 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026854s I0418 14:28:46.993870 138541847643968 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 50.794 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03036952018737793 s) (per-host) I0418 14:28:46.993922 138541847643968 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030433s (batch_requests_ready=0.001752s, total_serialization_initiated=0.028612s, others=0.000069s) I0418 14:28:46.994016 138541847643968 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034707s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.034696) I0418 14:28:46.994608 138541847643968 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.17 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/9. I0418 14:28:46.994935 138408567858944 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:48:46.994897 I0418 14:28:47.004135 138541847643968 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:28:47.004387 138412904412928 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:28:47.004535 138541847643968 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776522500.817775, 'wait_for_prev_duration_secs': 25.00061297416687, 'time_between_consecutive_saves_sec': 24.652878761291504, 'checkpointer_blocking_start_time': 1776522525.8197083, 'checkpointer_blocking_duration_secs': 1.1753959655761719, 'get_old_steps_start_time': 1776522526.99512, 'get_old_steps_duration_secs': 2.0742416381835938e-05, 'checkpoint_manager_blocking_start_time': 1776522500.815916, 'checkpoint_manager_blocking_duration_secs': 26.18858814239502} I0418 14:28:47.004635 138541847643968 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 14:28:47.004677 138541847643968 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:28:52.620837 138411830671104 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_16_zarr3_true/checkpoints/9/items/array_metadatas/process_7 I0418 14:29:29.728360 138408567858944 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 36.937 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.76482677459717 s) (per-host) I0418 14:29:29.728467 138408567858944 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.733414s. I0418 14:29:36.503371 138408567858944 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 49.508301s. I0418 14:29:36.503655 138412904412928 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:29:36.503774 138412904412928 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:29:36.503822 138412904412928 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:29:36.506523 138412904412928 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 14:29:36.506704 138541847643968 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 14:29:36.506850 138541847643968 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:29:36.507823 138541847643968 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.186, Tokens/s/device: 13895.202, total_weights: 65536, loss: 7.181, lm_loss: 7.181, perplexity: 1314.495 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sat Apr 18 14:29:47 UTC 2026 EXIT_CODE=0