XPK Start: Fri Apr 17 02:24:06 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 02:24:30.217639: 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 02:24:30.396887 134049321228096 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:24:39.435559 134049321228096 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:24:39.437929 134049321228096 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-sv-496n4-slice-job-0-0.mt-16-zarr3-false-sv-496n4:8482 I0417 02:24:40.441450 134049321228096 max_utils.py:284] Jax distributed system initialized! I0417 02:24:46.434564 134049321228096 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:24:46.434668 134049321228096 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:24:46.434707 134049321228096 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 02:24:46.434741 134049321228096 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:24:47.095662 134049321228096 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:24:47.208381 134049321228096 checkpointing.py:688] Setting up checkpoint logger... I0417 02:24:47.208492 134049321228096 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0417 02:24:47.208539 134049321228096 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:24:47.208745 134049321228096 base_pytree_checkpoint_handler.py:441] 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 0x79ea33b1caa0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:24:50.218025 134049321228096 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:24:50.218261 134049321228096 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79d6484850d0>}, handler_registry=None I0417 02:24:50.218493 134049321228096 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79d6484850d0>` 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 02:24:50.218540 134049321228096 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79d620666960>` 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 02:24:50.218575 134049321228096 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 0x79d6484850d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79d6484850d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79d620666960>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79d620666960>}). I0417 02:24:50.218902 134049321228096 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:24:50.218972 134049321228096 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x79d648292020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:24:52.207983 134049321228096 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints I0417 02:24:52.224588 134049321228096 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79d6486c5670> I0417 02:24:52.224711 134049321228096 checkpointing.py:302] Checkpoint manager created! I0417 02:24:53.189177 134049321228096 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:24:53.189296 134049321228096 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 02:24:54.546924 134049321228096 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:24:54.547040 134049321228096 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:24:54.547092 134049321228096 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:24:54.547134 134049321228096 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:24:54.547172 134049321228096 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:24:54.547206 134049321228096 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:24:54.547241 134049321228096 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:24:54.547275 134049321228096 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:24:54.547308 134049321228096 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:24:54.547341 134049321228096 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:24:54.547373 134049321228096 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:24:54.547403 134049321228096 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:24:54.946879 134049321228096 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:24:54.946970 134049321228096 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:24:54.952363 134049321228096 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:24:54.952421 134049321228096 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:24:54.969058 134049321228096 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:24:54.969123 134049321228096 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:24:54.985211 134049321228096 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:24:54.985265 134049321228096 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:24:55.009675 134049321228096 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 02:24:55.009738 134049321228096 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:24:55.025925 134049321228096 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 02:24:55.025985 134049321228096 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:24:55.042108 134049321228096 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 02:24:55.042169 134049321228096 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:24:55.072397 134049321228096 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:24:55.072468 134049321228096 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:24:55.092513 134049321228096 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:24:55.092576 134049321228096 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:25:06.098304 134049321228096 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 02:25:06.107428 134049321228096 metric_logger.py:289] number parameters: 1.104 billion I0417 02:25:17.817569 134049321228096 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 02:25:17.969889 134049321228096 checkpointing.py:798] Waited 0.15228939056396484 seconds for step 0 to finish before starting checkpointing. I0417 02:25:17.972660 134049321228096 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:25:17.974349 134049321228096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 02:25:17.975864 134049321228096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0. I0417 02:25:18.693422 134049321228096 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:25:18.712812 134049321228096 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:25:18.712906 134049321228096 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 02:25:18.975322 133921710798592 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0 I0417 02:25:18.990809 134049321228096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.279466s I0417 02:25:18.991565 134049321228096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.313 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2903287410736084 s) (per-host) I0417 02:25:18.991640 134049321228096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290423s (batch_requests_ready=0.003063s, total_serialization_initiated=0.286699s, others=0.000661s) I0417 02:25:18.991724 134049321228096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294329s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.294312) I0417 02:25:18.992518 134049321228096 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0. I0417 02:25:18.992803 133922190972672 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:45:18.992771 I0417 02:25:18.997021 134049321228096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:25:18.997322 133920042137344 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:25:18.997475 134049321228096 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392717.972642, 'wait_for_prev_duration_secs': 6.270408630371094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392717.9743876, 'checkpointer_blocking_duration_secs': 1.0185818672180176, 'get_old_steps_start_time': 1776392718.9929957, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776392717.9702783, 'checkpoint_manager_blocking_duration_secs': 1.0271592140197754} I0417 02:25:18.997579 134049321228096 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 02:25:18.997649 134049321228096 max_utils.py:750] Memstats: After params initialized: I0417 02:25:18.997698 134049321228096 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 02:25:18.997729 134049321228096 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 02:25:18.997754 134049321228096 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 02:25:18.997777 134049321228096 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 02:25:19.339061 134049321228096 metric_logger.py:185] completed step: 0, seconds: 11.412, TFLOP/s/device: 1.191, Tokens/s/device: 179.454, total_weights: 65536, loss: 10.874 I0417 02:25:19.342343 134049321228096 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/tensorboard/' I0417 02:25:19.741284 134049321228096 metric_logger.py:185] completed step: 1, seconds: 1.518, TFLOP/s/device: 8.950, Tokens/s/device: 1349.023, total_weights: 65536, loss: 10.874 I0417 02:25:19.883498 134049321228096 metric_logger.py:185] completed step: 2, seconds: 0.403, TFLOP/s/device: 33.696, Tokens/s/device: 5079.038, total_weights: 65536, loss: 10.560 I0417 02:25:20.030885 134049321228096 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1124.669, Tokens/s/device: 169522.391, total_weights: 65536, loss: 9.979 I0417 02:25:20.042909 134049321228096 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0417 02:25:20.325906 134049321228096 checkpointing.py:798] Waited 0.2829403877258301 seconds for step 5 to finish before starting checkpointing. I0417 02:25:20.328653 134049321228096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:25:20.746860 133921694013184 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/items I0417 02:25:20.910225 133919548036864 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392720494791120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776392720494791120, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 02:25:21.912229 2797 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:25:23.695636 133920058922752 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0417 02:26:03.285442 133920050530048 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.388232s (commit=39.756397s, array_metadata_write=1.631835s) I0417 02:26:03.287000 133922190972672 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.428 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.58562469482422 s) (per-host) I0417 02:26:03.287266 133922190972672 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.294322s. I0417 02:26:04.097286 133922190972672 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392720494791120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 02:26:04.857011 133922190972672 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000744s. I0417 02:26:04.872437 133919548036864 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392720494791120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 02:26:06.437771 133922190972672 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.210775s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/items I0417 02:26:06.439895 133922190972672 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0/items I0417 02:26:06.976156 133922190972672 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0 I0417 02:26:08.384035 133922190972672 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0`. I0417 02:26:08.385188 133922190972672 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.41 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/0 I0417 02:26:08.386895 133922190972672 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.393951s. I0417 02:26:08.387101 133920042137344 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:26:08.387206 133920042137344 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:26:08.387310 133920042137344 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:26:08.389747 133920042137344 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 02:26:08.390082 134049321228096 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0417 02:26:08.390263 134049321228096 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.061612 seconds. If this number is high, consider checkpointing less frequently. I0417 02:26:08.392117 134049321228096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0417 02:26:08.394500 134049321228096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5. I0417 02:26:09.136959 134049321228096 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:26:09.137089 134049321228096 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 02:26:09.176966 134049321228096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042979s I0417 02:26:09.178000 134049321228096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 24.748 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0623326301574707 s) (per-host) I0417 02:26:09.178104 134049321228096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.062459s (batch_requests_ready=0.006323s, total_serialization_initiated=0.055222s, others=0.000914s) I0417 02:26:09.178241 134049321228096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.067449s (all_items=0.000029s, per_item={'items': '0.00002933'}, temp_paths=0.067420) I0417 02:26:09.179219 134049321228096 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.79 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5. I0417 02:26:09.179595 133919009658624 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:46:09.179556 I0417 02:26:09.182080 134049321228096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:26:09.182397 133920042137344 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:26:09.182584 134049321228096 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392720.3286078, 'wait_for_prev_duration_secs': 48.061612129211426, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392768.3921869, 'checkpointer_blocking_duration_secs': 0.787585973739624, 'get_old_steps_start_time': 1776392769.1797998, 'get_old_steps_duration_secs': 5.984306335449219e-05, 'checkpoint_manager_blocking_start_time': 1776392720.3263757, 'checkpoint_manager_blocking_duration_secs': 48.856154680252075} I0417 02:26:09.182735 134049321228096 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0417 02:26:09.349580 133922190972672 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5 I0417 02:26:11.963614 133921165534976 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5/items I0417 02:26:14.833090 133919026444032 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0417 02:26:32.826958 134049321228096 metric_logger.py:185] completed step: 4, seconds: 0.144, TFLOP/s/device: 94.582, Tokens/s/device: 14256.477, total_weights: 65536, loss: 9.457 I0417 02:26:32.837521 134049321228096 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.644, Tokens/s/device: 13964.366, total_weights: 65536, loss: 8.955 I0417 02:26:32.981544 134049321228096 metric_logger.py:185] completed step: 6, seconds: 72.794, TFLOP/s/device: 0.187, Tokens/s/device: 28.134, total_weights: 65536, loss: 8.465 I0417 02:26:33.128907 134049321228096 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1621.763, Tokens/s/device: 244449.749, total_weights: 65536, loss: 7.998 I0417 02:26:33.276563 134049321228096 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.278, Tokens/s/device: 14059.850, total_weights: 65536, loss: 7.567 I0417 02:26:33.424902 134049321228096 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 02:26:33.425591 134049321228096 checkpointing.py:798] Waited 0.0006999969482421875 seconds for step 9 to finish before starting checkpointing. I0417 02:26:33.427876 134049321228096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:26:54.588147 133919018051328 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.513783s (commit=40.327217s, array_metadata_write=1.186565s) I0417 02:26:54.589660 133919009658624 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.736 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.47397255897522 s) (per-host) I0417 02:26:54.589789 133919009658624 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.410051s. I0417 02:26:56.120395 133919009658624 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000399s. I0417 02:26:57.599680 133919009658624 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.187872s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5/items I0417 02:26:57.601499 133919009658624 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5/items I0417 02:26:58.578513 133919009658624 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5 I0417 02:27:00.472007 133919009658624 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5`. I0417 02:27:00.472868 133919009658624 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.08 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/5 I0417 02:27:00.474385 133919009658624 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.294646s. I0417 02:27:00.474597 133920042137344 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:27:00.474716 133920042137344 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:27:00.474783 133920042137344 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:27:00.477581 133920042137344 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 02:27:00.477721 134049321228096 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0417 02:27:00.477797 134049321228096 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.049974 seconds. If this number is high, consider checkpointing less frequently. I0417 02:27:00.479596 134049321228096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 02:27:00.481638 134049321228096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9. I0417 02:27:01.189016 134049321228096 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:27:01.189117 134049321228096 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 02:27:01.215775 134049321228096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028133s I0417 02:27:01.216617 134049321228096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 42.053 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.036681413650512695 s) (per-host) I0417 02:27:01.216685 134049321228096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.036767s (batch_requests_ready=0.002595s, total_serialization_initiated=0.033396s, others=0.000775s) I0417 02:27:01.216756 134049321228096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040734s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.040718) I0417 02:27:01.217519 134049321228096 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9. I0417 02:27:01.217877 133921677227776 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:47:01.217815 I0417 02:27:01.224412 134049321228096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:27:01.224700 133920042137344 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:27:01.224828 134049321228096 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392793.427801, 'wait_for_prev_duration_secs': 27.049973726272583, 'time_between_consecutive_saves_sec': 25.03798794746399, 'checkpointer_blocking_start_time': 1776392820.4796808, 'checkpointer_blocking_duration_secs': 0.7383456230163574, 'get_old_steps_start_time': 1776392821.218047, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776392793.4258173, 'checkpoint_manager_blocking_duration_secs': 27.798978328704834} I0417 02:27:01.224975 134049321228096 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 02:27:01.225018 134049321228096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:27:01.418118 133919009658624 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9 I0417 02:27:03.215265 133923271055104 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9/items I0417 02:27:06.674202 133922190972672 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0417 02:27:46.813051 133921685620480 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.306721s (commit=41.129362s, array_metadata_write=1.177359s) I0417 02:27:46.814162 133921677227776 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.614 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.63420581817627 s) (per-host) I0417 02:27:46.814276 133921677227776 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.596276s. I0417 02:27:48.604649 133921677227776 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000404s. I0417 02:27:49.201203 133921677227776 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.466050s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9/items I0417 02:27:49.202787 133921677227776 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9/items I0417 02:27:49.736011 133921677227776 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9 I0417 02:27:51.155380 133921677227776 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9`. I0417 02:27:51.156176 133921677227776 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.68 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_false/checkpoints/9 I0417 02:27:51.157583 133921677227776 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.939579s. I0417 02:27:51.157772 133920042137344 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:27:51.157895 133920042137344 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:27:51.157948 133920042137344 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:27:51.159596 133920042137344 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 02:27:51.159760 134049321228096 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 02:27:51.159952 134049321228096 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:27:51.160629 134049321228096 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.326, Tokens/s/device: 13916.447, 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 02:28:01 UTC 2026 EXIT_CODE=0