XPK Start: Fri Apr 17 06:17:52 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 06:18:16.814483: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0417 06:18:16.991810 137732265080640 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 06:18:26.031355 137732265080640 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 06:18:26.033962 137732265080640 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-0dg85-slice-job-0-0.mt-16-zarr3-true-sv-0dg85:8482 I0417 06:18:27.846153 137732265080640 max_utils.py:284] Jax distributed system initialized! I0417 06:18:33.771746 137732265080640 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 06:18:33.771848 137732265080640 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 06:18:33.771888 137732265080640 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0417 06:18:33.771923 137732265080640 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 06:18:34.458995 137732265080640 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 06:18:34.574761 137732265080640 checkpointing.py:688] Setting up checkpoint logger... I0417 06:18:34.574876 137732265080640 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 06:18:34.574922 137732265080640 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 06:18:34.575131 137732265080640 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 0x7d43b6def740>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 06:18:37.833832 137732265080640 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 06:18:37.834070 137732265080640 checkpoint_manager.py:708] [process=1][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2e4c3b6e10>}, handler_registry=None I0417 06:18:37.834305 137732265080640 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2e4c3b6e10>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0417 06:18:37.834352 137732265080640 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2e4c737290>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0417 06:18:37.834387 137732265080640 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 0x7d2e4c3b6e10>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2e4c3b6e10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2e4c737290>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2e4c737290>}). I0417 06:18:37.834720 137732265080640 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 06:18:37.834792 137732265080640 async_checkpointer.py:192] [process=1][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d2e4c786020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 06:18:38.948158 137732265080640 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints I0417 06:18:38.950371 137732265080640 checkpoint_manager.py:929] [process=1][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/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d2f141c4920> I0417 06:18:38.950497 137732265080640 checkpointing.py:302] Checkpoint manager created! I0417 06:18:39.912131 137732265080640 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 06:18:39.912244 137732265080640 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 06:18:41.348705 137732265080640 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 06:18:41.348824 137732265080640 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:18:41.348897 137732265080640 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:18:41.348961 137732265080640 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 06:18:41.349008 137732265080640 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:18:41.349066 137732265080640 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:18:41.349109 137732265080640 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:18:41.349165 137732265080640 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 06:18:41.349213 137732265080640 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:18:41.349270 137732265080640 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:18:41.349320 137732265080640 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 06:18:41.349369 137732265080640 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 06:18:41.750310 137732265080640 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:18:41.750401 137732265080640 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:18:41.755643 137732265080640 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:18:41.755698 137732265080640 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:18:41.771728 137732265080640 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:18:41.771786 137732265080640 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:18:41.787329 137732265080640 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:18:41.787389 137732265080640 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:18:41.811111 137732265080640 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:18:41.811177 137732265080640 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:18:41.826762 137732265080640 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:18:41.826821 137732265080640 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:18:41.842429 137732265080640 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:18:41.842499 137732265080640 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:18:41.872135 137732265080640 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 06:18:41.872206 137732265080640 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:18:41.891917 137732265080640 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 06:18:41.891979 137732265080640 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 06:18:52.911447 137732265080640 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. I0417 06:18:52.913683 137732265080640 metric_logger.py:289] number parameters: 1.104 billion I0417 06:19:04.012833 137732265080640 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 06:19:04.164563 137732265080640 checkpointing.py:798] Waited 0.1517188549041748 seconds for step 0 to finish before starting checkpointing. I0417 06:19:04.167220 137732265080640 checkpoint_manager.py:2009] [process=1][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:19:04.168823 137732265080640 checkpoint_manager.py:1512] [process=1] Saving checkpoint at step 0 I0417 06:19:04.170402 137732265080640 event_tracking.py:70] [process=1] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/0. I0417 06:19:04.467397 137732265080640 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 06:19:04.468210 137732265080640 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:19:04.468269 137732265080640 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:19:04.739090 137732265080640 base_pytree_checkpoint_handler.py:154] [process=1][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272460s I0417 06:19:04.739263 137732265080640 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/blocking_gbytes_per_sec: 5.474 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28178954124450684 s) (per-host) I0417 06:19:04.739318 137732265080640 base_pytree_checkpoint_handler.py:768] [process=1][thread=MainThread] Initiated Pytree async_save. Time taken: 0.281854s (batch_requests_ready=0.003212s, total_serialization_initiated=0.278570s, others=0.000072s) I0417 06:19:04.739419 137732265080640 composite_checkpoint_handler.py:715] [process=1][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.286204s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.286187) I0417 06:19:04.740249 137732265080640 event_tracking.py:125] [process=1] [async] Finished blocking save in 0.57 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/0. I0417 06:19:04.740562 137602485548800 async_checkpointer.py:76] [process=1][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:39:04.740529 I0417 06:19:04.754168 137732265080640 checkpoint_manager.py:1560] [process=1][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:19:04.754420 137601951852288 async_checkpointer.py:280] [process=1][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:19:04.754578 137732265080640 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406744.1672015, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406744.1688626, 'checkpointer_blocking_duration_secs': 0.5718467235565186, 'get_old_steps_start_time': 1776406744.7407353, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776406744.1649735, 'checkpoint_manager_blocking_duration_secs': 0.5895614624023438} I0417 06:19:04.754695 137732265080640 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 06:19:04.754769 137732265080640 max_utils.py:750] Memstats: After params initialized: I0417 06:19:04.754821 137732265080640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_2(process=1,(2,0,0,0)) I0417 06:19:04.754854 137732265080640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_3(process=1,(3,0,0,0)) I0417 06:19:04.754881 137732265080640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_6(process=1,(2,1,0,0)) I0417 06:19:04.754905 137732265080640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_7(process=1,(3,1,0,0)) I0417 06:19:05.066813 137732265080640 metric_logger.py:185] completed step: 0, seconds: 11.098, TFLOP/s/device: 1.224, Tokens/s/device: 184.536, total_weights: 65536, loss: 10.874 I0417 06:19:05.239865 137732265080640 metric_logger.py:185] completed step: 1, seconds: 1.052, TFLOP/s/device: 12.912, Tokens/s/device: 1946.206, total_weights: 65536, loss: 10.874 I0417 06:19:05.687829 137732265080640 metric_logger.py:185] completed step: 2, seconds: 0.027, TFLOP/s/device: 504.273, Tokens/s/device: 76009.501, total_weights: 65536, loss: 10.560 I0417 06:19:05.835319 137732265080640 metric_logger.py:185] completed step: 3, seconds: 0.438, TFLOP/s/device: 31.037, Tokens/s/device: 4678.288, total_weights: 65536, loss: 9.979 I0417 06:19:05.840887 137732265080640 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0417 06:19:06.130394 137732265080640 checkpointing.py:798] Waited 0.2894766330718994 seconds for step 5 to finish before starting checkpointing. I0417 06:19:06.133283 137732265080640 checkpoint_manager.py:2020] [process=1][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:19:07.712242 2795 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 06:19:09.734092 137601960244992 array_metadata_store.py:203] [process=1][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/0/items/array_metadatas/process_1 I0417 06:19:41.006754 137602485548800 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/gbytes_per_sec: 43.218 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.54924273490906 s) (per-host) I0417 06:19:41.006877 137602485548800 async_checkpointer.py:90] [process=1][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.266203s. I0417 06:19:49.478987 137602485548800 async_checkpointer.py:160] [process=1][thread=async_save] Background save thread done. Time taken: 44.738296s. I0417 06:19:49.479236 137601951852288 async_checkpointer.py:288] [process=1][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:19:49.479295 137601951852288 async_checkpointer.py:298] [process=1][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:19:49.479342 137601951852288 checkpoint_manager.py:2137] [process=1][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:19:49.481211 137601951852288 checkpoint_manager.py:2146] [process=1][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 06:19:49.481398 137732265080640 checkpoint_manager.py:2032] [process=1][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0417 06:19:49.481559 137732265080640 checkpoint_manager.py:1452] Waiting for previous save to complete took 43.348274 seconds. If this number is high, consider checkpointing less frequently. I0417 06:19:49.483523 137732265080640 checkpoint_manager.py:1512] [process=1] Saving checkpoint at step 5 I0417 06:19:49.485889 137732265080640 event_tracking.py:70] [process=1] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/5. I0417 06:19:49.812756 137732265080640 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:19:49.812850 137732265080640 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:19:49.848146 137732265080640 base_pytree_checkpoint_handler.py:154] [process=1][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.036952s I0417 06:19:49.848283 137732265080640 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/blocking_gbytes_per_sec: 35.002 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04407167434692383 s) (per-host) I0417 06:19:49.848330 137732265080640 base_pytree_checkpoint_handler.py:768] [process=1][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044128s (batch_requests_ready=0.002809s, total_serialization_initiated=0.041256s, others=0.000063s) I0417 06:19:49.848412 137732265080640 composite_checkpoint_handler.py:715] [process=1][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049046s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.049030) I0417 06:19:49.849139 137732265080640 event_tracking.py:125] [process=1] [async] Finished blocking save in 0.37 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/5. I0417 06:19:49.849493 137601951852288 async_checkpointer.py:76] [process=1][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:39:49.849444 I0417 06:19:49.868991 137732265080640 checkpoint_manager.py:1560] [process=1][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:19:49.869315 137581643675392 async_checkpointer.py:280] [process=1][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:19:49.869488 137732265080640 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406746.1332464, 'wait_for_prev_duration_secs': 43.34827399253845, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406789.4835618, 'checkpointer_blocking_duration_secs': 0.36607837677001953, 'get_old_steps_start_time': 1776406789.849667, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776406746.1307693, 'checkpoint_manager_blocking_duration_secs': 43.73867201805115} I0417 06:19:49.869598 137732265080640 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0417 06:19:49.870470 137732265080640 metric_logger.py:185] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.873, Tokens/s/device: 12491.461, total_weights: 65536, loss: 9.457 I0417 06:19:49.877947 137732265080640 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.406, Tokens/s/device: 13928.467, total_weights: 65536, loss: 8.955 I0417 06:19:55.015800 137601960244992 array_metadata_store.py:203] [process=1][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/5/items/array_metadatas/process_1 I0417 06:20:13.045689 137732265080640 metric_logger.py:185] completed step: 6, seconds: 44.035, TFLOP/s/device: 0.309, Tokens/s/device: 46.508, total_weights: 65536, loss: 8.465 I0417 06:20:13.193305 137732265080640 metric_logger.py:185] completed step: 7, seconds: 23.022, TFLOP/s/device: 0.590, Tokens/s/device: 88.957, total_weights: 65536, loss: 7.998 I0417 06:20:13.340779 137732265080640 metric_logger.py:185] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.799, Tokens/s/device: 13384.746, total_weights: 65536, loss: 7.567 I0417 06:20:13.489286 137732265080640 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 06:20:13.489982 137732265080640 checkpointing.py:798] Waited 0.0007002353668212891 seconds for step 9 to finish before starting checkpointing. I0417 06:20:13.492052 137732265080640 checkpoint_manager.py:2020] [process=1][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:20:30.822082 137601951852288 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/gbytes_per_sec: 38.510 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.01783061027527 s) (per-host) I0417 06:20:30.822210 137601951852288 async_checkpointer.py:90] [process=1][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.972600s. I0417 06:20:39.568562 137601951852288 async_checkpointer.py:160] [process=1][thread=async_save] Background save thread done. Time taken: 49.718937s. I0417 06:20:39.568857 137581643675392 async_checkpointer.py:288] [process=1][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:20:39.568989 137581643675392 async_checkpointer.py:298] [process=1][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:20:39.569042 137581643675392 checkpoint_manager.py:2137] [process=1][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:20:39.571703 137581643675392 checkpoint_manager.py:2146] [process=1][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 06:20:39.571882 137732265080640 checkpoint_manager.py:2032] [process=1][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0417 06:20:39.572002 137732265080640 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.079965 seconds. If this number is high, consider checkpointing less frequently. I0417 06:20:39.573537 137732265080640 checkpoint_manager.py:1512] [process=1] Saving checkpoint at step 9 I0417 06:20:39.576001 137732265080640 event_tracking.py:70] [process=1] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/9. I0417 06:20:40.312000 137732265080640 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:20:40.312097 137732265080640 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:20:40.338856 137732265080640 base_pytree_checkpoint_handler.py:154] [process=1][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028430s I0417 06:20:40.339023 137732265080640 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/blocking_gbytes_per_sec: 43.262 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03565645217895508 s) (per-host) I0417 06:20:40.339075 137732265080640 base_pytree_checkpoint_handler.py:768] [process=1][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035719s (batch_requests_ready=0.002993s, total_serialization_initiated=0.032659s, others=0.000068s) I0417 06:20:40.339177 137732265080640 composite_checkpoint_handler.py:715] [process=1][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.039881s (all_items=0.000010s, per_item={'items': '0.00001001'}, temp_paths=0.039871) I0417 06:20:40.339782 137732265080640 event_tracking.py:125] [process=1] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/9. I0417 06:20:40.340074 137599234582272 async_checkpointer.py:76] [process=1][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:40:40.340037 I0417 06:20:40.356758 137732265080640 checkpoint_manager.py:1560] [process=1][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:20:40.357047 137581643675392 async_checkpointer.py:280] [process=1][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:20:40.357187 137732265080640 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406813.4920146, 'wait_for_prev_duration_secs': 26.079964637756348, 'time_between_consecutive_saves_sec': 24.010761976242065, 'checkpointer_blocking_start_time': 1776406839.5735745, 'checkpointer_blocking_duration_secs': 0.766646146774292, 'get_old_steps_start_time': 1776406840.340236, 'get_old_steps_duration_secs': 2.193450927734375e-05, 'checkpoint_manager_blocking_start_time': 1776406813.4902265, 'checkpoint_manager_blocking_duration_secs': 26.86692976951599} I0417 06:20:40.357359 137732265080640 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 06:20:40.357403 137732265080640 checkpoint_manager.py:2020] [process=1][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:20:45.394553 137601960244992 array_metadata_store.py:203] [process=1][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_16_zarr3_true/checkpoints/9/items/array_metadatas/process_1 I0417 06:21:21.151490 137599234582272 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/gbytes_per_sec: 38.670 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.848079204559326 s) (per-host) I0417 06:21:21.151604 137599234582272 async_checkpointer.py:90] [process=1][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.811419s. I0417 06:21:30.008600 137599234582272 async_checkpointer.py:160] [process=1][thread=async_save] Background save thread done. Time taken: 49.668398s. I0417 06:21:30.008917 137581643675392 async_checkpointer.py:288] [process=1][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:21:30.009039 137581643675392 async_checkpointer.py:298] [process=1][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:21:30.009102 137581643675392 checkpoint_manager.py:2137] [process=1][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:21:30.011795 137581643675392 checkpoint_manager.py:2146] [process=1][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 06:21:30.011968 137732265080640 checkpoint_manager.py:2032] [process=1][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 06:21:30.012134 137732265080640 checkpoint_manager.py:2009] [process=1][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:21:30.012794 137732265080640 metric_logger.py:185] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.819, Tokens/s/device: 13839.895, total_weights: 65536, loss: 7.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 06:21:42 UTC 2026 EXIT_CODE=0