XPK Start: Fri Apr 17 05:30:45 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 05:31:09.300571: 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 05:31:09.479948 137711000848192 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 05:31:18.520063 137711000848192 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 05:31:18.522292 137711000848192 distributed.py:172] Connecting to JAX distributed service on mt-11-optimizer-offload-gwo6u-slice-job-0-0.mt-11-optimizer-offload-gwo6u:8482 I0417 05:31:20.113451 137711000848192 max_utils.py:284] Jax distributed system initialized! I0417 05:31:26.335514 137711000848192 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 05:31:26.335613 137711000848192 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 05:31:26.335652 137711000848192 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 05:31:26.335687 137711000848192 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 05:31:26.994264 137711000848192 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 05:31:27.108608 137711000848192 checkpointing.py:688] Setting up checkpoint logger... I0417 05:31:27.108720 137711000848192 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 05:31:27.108765 137711000848192 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 05:31:27.108975 137711000848192 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 0x7d3ebbf14b60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 05:31:30.110269 137711000848192 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 05:31:30.110505 137711000848192 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 0x7d2a742735f0>}, handler_registry=None I0417 05:31:30.110739 137711000848192 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2a742735f0>` 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 05:31:30.110786 137711000848192 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d294c552360>` 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 05:31:30.110821 137711000848192 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 0x7d2a742735f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2a742735f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d294c552360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d294c552360>}). I0417 05:31:30.111158 137711000848192 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 05:31:30.111229 137711000848192 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d2a3c326020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 05:31:32.122997 137711000848192 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_11_optimizer_offload_false/checkpoints I0417 05:31:32.125228 137711000848192 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d2a74286300> I0417 05:31:32.125340 137711000848192 checkpointing.py:302] Checkpoint manager created! I0417 05:31:33.076539 137711000848192 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 05:31:33.076653 137711000848192 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 05:31:34.456393 137711000848192 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 05:31:34.456509 137711000848192 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 05:31:34.456562 137711000848192 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 05:31:34.456605 137711000848192 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 05:31:34.456642 137711000848192 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 05:31:34.456676 137711000848192 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 05:31:34.456712 137711000848192 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 05:31:34.456746 137711000848192 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 05:31:34.456779 137711000848192 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 05:31:34.456812 137711000848192 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 05:31:34.456861 137711000848192 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 05:31:34.456895 137711000848192 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 05:31:34.859018 137711000848192 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 05:31:34.859108 137711000848192 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 05:31:34.864448 137711000848192 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 05:31:34.864503 137711000848192 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 05:31:34.880950 137711000848192 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 05:31:34.881020 137711000848192 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 05:31:34.896748 137711000848192 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 05:31:34.896807 137711000848192 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 05:31:34.920912 137711000848192 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 05:31:34.920977 137711000848192 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 05:31:34.936792 137711000848192 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 05:31:34.936864 137711000848192 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 05:31:34.952604 137711000848192 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 05:31:34.952662 137711000848192 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 05:31:34.988866 137711000848192 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 05:31:34.988967 137711000848192 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 05:31:35.009324 137711000848192 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 05:31:35.009388 137711000848192 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 05:31:45.892303 137711000848192 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 05:31:45.901014 137711000848192 metric_logger.py:289] number parameters: 1.104 billion I0417 05:31:57.407164 137711000848192 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 05:31:57.558747 137711000848192 checkpointing.py:798] Waited 0.15157341957092285 seconds for step 0 to finish before starting checkpointing. I0417 05:31:57.561125 137711000848192 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 05:31:57.562812 137711000848192 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 05:31:57.564341 137711000848192 event_tracking.py:70] [process=0] [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_11_optimizer_offload_false/checkpoints/0. I0417 05:31:58.127576 137711000848192 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 05:31:58.147009 137711000848192 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 05:31:58.147081 137711000848192 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 05:31:58.378290 137580486063872 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0 I0417 05:31:58.420151 137711000848192 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274595s I0417 05:31:58.420930 137711000848192 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.404 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28542494773864746 s) (per-host) I0417 05:31:58.421013 137711000848192 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285532s (batch_requests_ready=0.003089s, total_serialization_initiated=0.281738s, others=0.000705s) I0417 05:31:58.421099 137711000848192 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289561s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.289544) I0417 05:31:58.421912 137711000848192 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.86 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_11_optimizer_offload_false/checkpoints/0. I0417 05:31:58.422248 137580966397696 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:51:58.422211 I0417 05:31:58.439172 137711000848192 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 05:31:58.439513 137576149509888 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 05:31:58.439670 137711000848192 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776403917.5611064, 'wait_for_prev_duration_secs': 6.747245788574219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776403917.5628633, 'checkpointer_blocking_duration_secs': 0.8595404624938965, 'get_old_steps_start_time': 1776403918.4224305, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776403917.559169, 'checkpoint_manager_blocking_duration_secs': 0.8804628849029541} I0417 05:31:58.439773 137711000848192 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 05:31:58.439865 137711000848192 max_utils.py:750] Memstats: After params initialized: I0417 05:31:58.439916 137711000848192 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 05:31:58.439956 137711000848192 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 05:31:58.439982 137711000848192 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 05:31:58.440007 137711000848192 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 05:31:58.794641 137711000848192 metric_logger.py:185] completed step: 0, seconds: 11.217, TFLOP/s/device: 1.211, Tokens/s/device: 182.582, total_weights: 65536, loss: 10.874 I0417 05:31:58.797793 137711000848192 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/tensorboard/' I0417 05:31:59.605762 137711000848192 metric_logger.py:185] completed step: 1, seconds: 1.384, TFLOP/s/device: 9.816, Tokens/s/device: 1479.620, total_weights: 65536, loss: 10.874 I0417 05:31:59.747485 137711000848192 metric_logger.py:185] completed step: 2, seconds: 0.812, TFLOP/s/device: 16.731, Tokens/s/device: 2521.885, total_weights: 65536, loss: 10.262 I0417 05:31:59.894966 137711000848192 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1120.496, Tokens/s/device: 168893.287, total_weights: 65536, loss: 9.732 I0417 05:32:00.613299 137580469278464 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/items I0417 05:32:00.763656 137576141117184 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776403920361077693, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776403920361077693, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 05:32:01.741876 2858 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 05:32:03.492946 137579895383808 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_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0417 05:32:23.277480 137711000848192 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.823, Tokens/s/device: 14292.693, total_weights: 65536, loss: 9.274 I0417 05:32:23.287387 137711000848192 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.701, Tokens/s/device: 13972.941, total_weights: 65536, loss: 8.888 I0417 05:32:23.431760 137711000848192 metric_logger.py:185] completed step: 6, seconds: 23.381, TFLOP/s/device: 0.581, Tokens/s/device: 87.593, total_weights: 65536, loss: 8.589 I0417 05:32:23.579274 137711000848192 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1790.371, Tokens/s/device: 269864.277, total_weights: 65536, loss: 8.380 I0417 05:32:23.726730 137711000848192 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.015, Tokens/s/device: 14020.195, total_weights: 65536, loss: 8.251 I0417 05:32:23.875164 137711000848192 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 05:32:23.875800 137711000848192 checkpointing.py:798] Waited 0.00064849853515625 seconds for step 9 to finish before starting checkpointing. I0417 05:32:23.877927 137711000848192 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 05:32:38.206833 137576698603264 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.462447s (commit=35.253360s, array_metadata_write=1.209088s) I0417 05:32:38.207950 137580966397696 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.419 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.07242250442505 s) (per-host) I0417 05:32:38.208073 137580966397696 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.785706s. I0417 05:32:38.965514 137580966397696 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776403920361077693, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 05:32:39.718909 137576141117184 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776403920361077693, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 05:32:39.942626 137580966397696 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 05:32:40.514483 137580966397696 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000419s. I0417 05:32:40.971410 137580966397696 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.881433s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/items I0417 05:32:40.973154 137580966397696 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0/items I0417 05:32:41.927419 137580966397696 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0 I0417 05:32:43.362421 137580966397696 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_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0`. I0417 05:32:43.363266 137580966397696 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 45.80 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/0 I0417 05:32:43.364708 137580966397696 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 44.942339s. I0417 05:32:43.364852 137576149509888 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 05:32:43.364916 137576149509888 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 05:32:43.364975 137576149509888 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 05:32:43.367648 137576149509888 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 05:32:43.367951 137711000848192 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 05:32:43.368094 137711000848192 checkpoint_manager.py:1452] Waiting for previous save to complete took 19.490167 seconds. If this number is high, consider checkpointing less frequently. I0417 05:32:43.369956 137711000848192 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 05:32:43.371937 137711000848192 event_tracking.py:70] [process=0] [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_11_optimizer_offload_false/checkpoints/9. I0417 05:32:44.090255 137711000848192 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 05:32:44.090352 137711000848192 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 05:32:44.118285 137711000848192 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029569s I0417 05:32:44.119065 137711000848192 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.155 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03841543197631836 s) (per-host) I0417 05:32:44.119134 137711000848192 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038501s (batch_requests_ready=0.002677s, total_serialization_initiated=0.035088s, others=0.000736s) I0417 05:32:44.119210 137711000848192 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042413s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.042396) I0417 05:32:44.119908 137711000848192 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 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_11_optimizer_offload_false/checkpoints/9. I0417 05:32:44.120317 137579363624704 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:52:44.120278 I0417 05:32:44.126133 137711000848192 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 05:32:44.126388 137576149509888 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 05:32:44.126521 137711000848192 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776403943.877893, 'wait_for_prev_duration_secs': 19.490167140960693, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776403963.3699996, 'checkpointer_blocking_duration_secs': 0.7504556179046631, 'get_old_steps_start_time': 1776403964.1204805, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776403943.876039, 'checkpoint_manager_blocking_duration_secs': 20.250447511672974} I0417 05:32:44.126631 137711000848192 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 05:32:44.126675 137711000848192 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 05:32:44.317057 137580966397696 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9 I0417 05:32:46.036579 137579412322048 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9/items I0417 05:32:49.534130 137579380410112 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_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0417 05:33:29.329094 137579372017408 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.980425s (commit=40.299512s, array_metadata_write=1.680913s) I0417 05:33:29.330362 137579363624704 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.909 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.24967622756958 s) (per-host) I0417 05:33:29.330487 137579363624704 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.210056s. I0417 05:33:31.002645 137579363624704 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 05:33:31.166611 137579363624704 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000391s. I0417 05:33:31.933525 137579363624704 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.738316s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9/items I0417 05:33:31.935058 137579363624704 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9/items I0417 05:33:32.648999 137579363624704 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9 I0417 05:33:34.485942 137579363624704 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_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9`. I0417 05:33:34.486753 137579363624704 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.12 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_11_optimizer_offload_false/checkpoints/9 I0417 05:33:34.487970 137579363624704 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.367537s. I0417 05:33:34.488116 137576149509888 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 05:33:34.488176 137576149509888 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 05:33:34.488230 137576149509888 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 05:33:34.490132 137576149509888 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 05:33:34.490302 137711000848192 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 05:33:34.490474 137711000848192 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 05:33:34.491259 137711000848192 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.502, Tokens/s/device: 13942.975, 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: Fri Apr 17 05:33:46 UTC 2026 EXIT_CODE=0