XPK Start: Thu Apr 16 22:10:40 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-16 22:11:04.421897: 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) I0416 22:11:04.947127 132188817147712 max_utils.py:273] Attempting to initialize the jax distributed system... I0416 22:11:13.985472 132188817147712 distributed.py:149] Starting JAX distributed service on [::]:8482 I0416 22:11:13.987787 132188817147712 distributed.py:172] Connecting to JAX distributed service on mt-01-base-3l924-slice-job-0-0.mt-01-base-3l924:8482 I0416 22:11:14.851120 132188817147712 max_utils.py:284] Jax distributed system initialized! I0416 22:11:20.767407 132188817147712 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 22:11:20.767508 132188817147712 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 22:11:20.767547 132188817147712 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 I0416 22:11:20.767580 132188817147712 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 22:11:21.429120 132188817147712 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 22:11:21.542934 132188817147712 checkpointing.py:688] Setting up checkpoint logger... I0416 22:11:21.543046 132188817147712 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 22:11:21.543090 132188817147712 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0416 22:11:21.543282 132188817147712 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 0x78390c9ef650>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 22:11:25.050117 132188817147712 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 22:11:25.050359 132188817147712 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 0x7824d46f0fe0>}, handler_registry=None I0416 22:11:25.050597 132188817147712 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7824d46f0fe0>` 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`. I0416 22:11:25.050644 132188817147712 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7824bc608b60>` 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`. I0416 22:11:25.050679 132188817147712 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 0x7824d46f0fe0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7824d46f0fe0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7824bc608b60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7824bc608b60>}). I0416 22:11:25.051026 132188817147712 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0416 22:11:25.051102 132188817147712 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7824d4389ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0416 22:11:26.293246 132188817147712 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints I0416 22:11:26.697232 132188817147712 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7824bc6b74a0> I0416 22:11:26.697400 132188817147712 checkpointing.py:302] Checkpoint manager created! I0416 22:11:27.632917 132188817147712 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 22:11:27.633035 132188817147712 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0416 22:11:28.999714 132188817147712 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 22:11:28.999830 132188817147712 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:11:28.999904 132188817147712 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 22:11:28.999948 132188817147712 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 22:11:28.999985 132188817147712 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:11:29.000020 132188817147712 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 22:11:29.000056 132188817147712 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:11:29.000096 132188817147712 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 22:11:29.000131 132188817147712 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:11:29.000164 132188817147712 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 22:11:29.000196 132188817147712 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 22:11:29.000227 132188817147712 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 22:11:29.405970 132188817147712 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:11:29.406063 132188817147712 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:11:29.411341 132188817147712 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:11:29.411396 132188817147712 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:11:29.427647 132188817147712 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:11:29.427712 132188817147712 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:11:29.443215 132188817147712 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:11:29.443271 132188817147712 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 22:11:29.467060 132188817147712 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:11:29.467132 132188817147712 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:11:29.482766 132188817147712 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:11:29.482826 132188817147712 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:11:29.498661 132188817147712 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:11:29.498721 132188817147712 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:11:29.528407 132188817147712 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 22:11:29.528476 132188817147712 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 22:11:29.548210 132188817147712 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 22:11:29.548275 132188817147712 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0416 22:11:40.666591 132188817147712 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. I0416 22:11:40.675599 132188817147712 metric_logger.py:289] number parameters: 1.104 billion I0416 22:11:51.959137 132188817147712 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 22:11:52.110687 132188817147712 checkpointing.py:798] Waited 0.1515357494354248 seconds for step 0 to finish before starting checkpointing. I0416 22:11:52.113054 132188817147712 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:11:52.114631 132188817147712 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0416 22:11:52.116022 132188817147712 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0. I0416 22:11:52.434896 132188817147712 signaling_client.py:364] Using JaxDistributedSignalingClient I0416 22:11:52.454495 132188817147712 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:11:52.454563 132188817147712 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 I0416 22:11:52.679759 132059742529280 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0 I0416 22:11:52.725330 132188817147712 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272326s I0416 22:11:52.726255 132188817147712 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.434 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28386688232421875 s) (per-host) I0416 22:11:52.726326 132188817147712 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283956s (batch_requests_ready=0.003376s, total_serialization_initiated=0.279738s, others=0.000842s) I0416 22:11:52.726408 132188817147712 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287899s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.287882) I0416 22:11:52.727164 132188817147712 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0. I0416 22:11:52.727443 132059750921984 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:31:52.727411 I0416 22:11:52.741402 132188817147712 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:11:52.741673 132057581733632 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:11:52.741803 132188817147712 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776377512.1130357, 'wait_for_prev_duration_secs': 6.794929504394531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776377512.1146703, 'checkpointer_blocking_duration_secs': 0.6129162311553955, 'get_old_steps_start_time': 1776377512.7276132, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776377512.111136, 'checkpoint_manager_blocking_duration_secs': 0.6306262016296387} I0416 22:11:52.741927 132188817147712 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 22:11:52.741997 132188817147712 max_utils.py:750] Memstats: After params initialized: I0416 22:11:52.742051 132188817147712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0416 22:11:52.742084 132188817147712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0416 22:11:52.742113 132188817147712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0416 22:11:52.742138 132188817147712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0416 22:11:53.076503 132188817147712 metric_logger.py:185] completed step: 0, seconds: 10.983, TFLOP/s/device: 1.237, Tokens/s/device: 186.469, total_weights: 65536, loss: 10.874 I0416 22:11:53.079793 132188817147712 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/tensorboard/' I0416 22:11:53.502168 132188817147712 metric_logger.py:185] completed step: 1, seconds: 1.114, TFLOP/s/device: 12.197, Tokens/s/device: 1838.458, total_weights: 65536, loss: 10.874 I0416 22:11:53.643781 132188817147712 metric_logger.py:185] completed step: 2, seconds: 0.427, TFLOP/s/device: 31.852, Tokens/s/device: 4801.088, total_weights: 65536, loss: 10.262 I0416 22:11:53.791424 132188817147712 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1121.235, Tokens/s/device: 169004.786, total_weights: 65536, loss: 9.732 I0416 22:11:54.499252 132059725743872 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/items I0416 22:11:54.612591 132055480399616 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377514201115904, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776377514201115904, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/_CHECKPOINT_METADATA I0416 22:11:55.869632 2785 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0416 22:11:57.969595 132057598519040 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/items/array_metadatas/process_0 I0416 22:12:16.247759 132188817147712 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.850, Tokens/s/device: 14296.784, total_weights: 65536, loss: 9.274 I0416 22:12:16.257665 132188817147712 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.363, Tokens/s/device: 13922.028, total_weights: 65536, loss: 8.888 I0416 22:12:16.402323 132188817147712 metric_logger.py:185] completed step: 6, seconds: 22.454, TFLOP/s/device: 0.605, Tokens/s/device: 91.207, total_weights: 65536, loss: 8.589 I0416 22:12:16.549716 132188817147712 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1781.919, Tokens/s/device: 268590.164, total_weights: 65536, loss: 8.380 I0416 22:12:16.697442 132188817147712 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 92.853, Tokens/s/device: 13995.763, total_weights: 65536, loss: 8.251 I0416 22:12:16.845918 132188817147712 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 22:12:16.846604 132188817147712 checkpointing.py:798] Waited 0.0006988048553466797 seconds for step 9 to finish before starting checkpointing. I0416 22:12:16.849006 132188817147712 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:12:32.068953 132057590126336 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.197013s (commit=34.946059s, array_metadata_write=1.250954s) I0416 22:12:32.070232 132059750921984 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.861 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 39.62779259681702 s) (per-host) I0416 22:12:32.070358 132059750921984 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.342804s. I0416 22:12:32.787603 132059750921984 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377514201115904, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/_CHECKPOINT_METADATA I0416 22:12:33.555412 132055480399616 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776377514201115904, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/_CHECKPOINT_METADATA I0416 22:12:33.626543 132059750921984 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000435s. I0416 22:12:34.163633 132059750921984 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:12:35.085205 132059750921984 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.169936s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/items I0416 22:12:35.086897 132059750921984 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0/items I0416 22:12:35.598170 132059750921984 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0 I0416 22:12:37.487323 132059750921984 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0`. I0416 22:12:37.488087 132059750921984 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 45.37 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/0 I0416 22:12:37.489393 132059750921984 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 44.761838s. I0416 22:12:37.489613 132057581733632 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:12:37.489735 132057581733632 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:12:37.489803 132057581733632 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:12:37.492606 132057581733632 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 22:12:37.492790 132188817147712 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. W0416 22:12:37.492960 132188817147712 checkpoint_manager.py:1452] Waiting for previous save to complete took 20.643953 seconds. If this number is high, consider checkpointing less frequently. I0416 22:12:37.494822 132188817147712 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0416 22:12:37.496760 132188817147712 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9. I0416 22:12:38.194382 132188817147712 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:12:38.194473 132188817147712 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 I0416 22:12:38.221939 132188817147712 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028974s I0416 22:12:38.222634 132188817147712 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.482 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03810524940490723 s) (per-host) I0416 22:12:38.222701 132188817147712 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038188s (batch_requests_ready=0.003031s, total_serialization_initiated=0.034509s, others=0.000648s) I0416 22:12:38.222773 132188817147712 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042074s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.042056) I0416 22:12:38.223506 132188817147712 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9. I0416 22:12:38.223858 132057581733632 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 22:32:38.223795 I0416 22:12:38.229706 132188817147712 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:12:38.229994 132058153305856 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:12:38.230156 132188817147712 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776377536.8489752, 'wait_for_prev_duration_secs': 20.64395260810852, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776377557.4948835, 'checkpointer_blocking_duration_secs': 0.7291295528411865, 'get_old_steps_start_time': 1776377558.2240415, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776377536.8468502, 'checkpoint_manager_blocking_duration_secs': 21.383270502090454} I0416 22:12:38.230269 132188817147712 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 22:12:38.230314 132188817147712 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:12:38.427949 132059750921984 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9 I0416 22:12:40.158101 132058693232384 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9/items I0416 22:12:43.032409 132058701625088 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9/items/array_metadatas/process_0 I0416 22:13:22.635817 132058161698560 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.360300s (commit=40.151170s, array_metadata_write=1.209130s) I0416 22:13:22.636997 132057581733632 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.535 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.452444076538086 s) (per-host) I0416 22:13:22.637117 132057581733632 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.413131s. I0416 22:13:24.512151 132057581733632 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000585s. I0416 22:13:25.169661 132057581733632 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:13:26.147627 132057581733632 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.211713s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9/items I0416 22:13:26.149351 132057581733632 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9/items I0416 22:13:26.682052 132057581733632 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9 I0416 22:13:28.099780 132057581733632 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9`. I0416 22:13:28.100584 132057581733632 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.61 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_01_base/checkpoints/9 I0416 22:13:28.102073 132057581733632 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.878085s. I0416 22:13:28.102285 132058153305856 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:13:28.102401 132058153305856 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:13:28.102462 132058153305856 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:13:28.104199 132058153305856 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 22:13:28.104320 132188817147712 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. I0416 22:13:28.104430 132188817147712 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:13:28.105121 132188817147712 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.357, Tokens/s/device: 13920.988, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 16 22:13:36 UTC 2026 EXIT_CODE=0