XPK Start: Wed Apr 22 15:15:00 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-22 15:15:24.198314: 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) I0422 15:15:24.738258 135141802145600 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 15:15:33.780191 135141802145600 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 15:15:33.782498 135141802145600 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-sapbf-slice-job-0-0.mt-15-ocdbt-true-rs-sapbf:8482 I0422 15:15:35.342771 135141802145600 max_utils.py:284] Jax distributed system initialized! I0422 15:15:41.594848 135141802145600 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 15:15:41.594953 135141802145600 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 15:15:41.594993 135141802145600 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 I0422 15:15:41.595028 135141802145600 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 15:15:42.292818 135141802145600 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 15:15:42.406260 135141802145600 checkpointing.py:688] Setting up checkpoint logger... I0422 15:15:42.406382 135141802145600 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 15:15:42.406426 135141802145600 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 15:15:42.406629 135141802145600 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 0x7ae892b1c3b0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 15:15:44.538444 135141802145600 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 15:15:44.538680 135141802145600 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ad43c673bf0>}, handler_registry=None I0422 15:15:44.538922 135141802145600 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ad43c673bf0>` 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`. I0422 15:15:44.538972 135141802145600 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ad46c1961b0>` 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`. I0422 15:15:44.539007 135141802145600 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 0x7ad43c673bf0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ad43c673bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ad46c1961b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ad46c1961b0>}). I0422 15:15:44.539373 135141802145600 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 15:15:44.539450 135141802145600 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ad4543cd9e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 15:15:46.091809 135020710700800 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776870781426064075, 'commit_timestamp_nsecs': 1776870828663485055, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0422 15:15:46.346374 135020710700800 checkpoint_manager.py:1871] Missing metrics for step 9 I0422 15:15:46.370061 135141802145600 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints I0422 15:15:46.769557 135141802145600 checkpoint_manager.py:929] [process=6][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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ad45432a840> I0422 15:15:46.769726 135141802145600 checkpointing.py:302] Checkpoint manager created! I0422 15:15:47.742183 135141802145600 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 15:15:47.742304 135141802145600 checkpointing.py:582] restoring from this run's directory step 9 I0422 15:15:47.987372 135141802145600 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9. I0422 15:15:48.239430 135141802145600 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9. I0422 15:15:48.239849 135141802145600 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9. I0422 15:15:49.683861 1448 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 15:15:57.538660 135141802145600 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 201.020 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.861278772354126 s) (per-host) I0422 15:15:57.546373 135141802145600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 190.558 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 8.305121660232544 s) (per-host) I0422 15:15:57.548654 135141802145600 event_tracking.py:138] [process=6] [sync] Finished load in 9.56 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9 I0422 15:15:57.548751 135141802145600 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776870947.9872746, 'checkpointer_duration_secs': 9.561448335647583, 'checkpoint_manager_start_time': 1776870947.7429152, 'checkpoint_manager_duration_secs': 9.805809736251831} fsdp: 32 I0422 15:15:57.624394 135141802145600 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0422 15:15:57.624507 135141802145600 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:15:57.624559 135141802145600 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:15:57.624601 135141802145600 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0422 15:15:57.624638 135141802145600 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:15:57.624671 135141802145600 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:15:57.624708 135141802145600 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:15:57.624743 135141802145600 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0422 15:15:57.624778 135141802145600 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:15:57.624810 135141802145600 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:15:57.624843 135141802145600 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0422 15:15:57.624875 135141802145600 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0422 15:15:58.026124 135141802145600 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:15:58.026215 135141802145600 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:15:58.031625 135141802145600 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:15:58.031680 135141802145600 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:15:58.048136 135141802145600 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:15:58.048196 135141802145600 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:15:58.063944 135141802145600 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:15:58.064002 135141802145600 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:15:58.087861 135141802145600 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:15:58.087930 135141802145600 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:15:58.103823 135141802145600 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:15:58.103886 135141802145600 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:15:58.119675 135141802145600 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:15:58.119734 135141802145600 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:15:58.149746 135141802145600 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 15:15:58.149816 135141802145600 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:15:58.169568 135141802145600 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 15:15:58.169630 135141802145600 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 15:16:09.753606 135141802145600 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. I0422 15:16:09.755985 135141802145600 metric_logger.py:301] number parameters: 1.104 billion I0422 15:16:20.883001 135141802145600 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0422 15:16:21.036071 135141802145600 checkpointing.py:798] Waited 0.15305614471435547 seconds for step 10 to finish before starting checkpointing. I0422 15:16:21.038345 135141802145600 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:16:21.040311 135141802145600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0422 15:16:21.041679 135141802145600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/10. I0422 15:16:21.389264 135141802145600 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 15:16:21.390404 135141802145600 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:16:21.390459 135141802145600 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 I0422 15:16:21.668655 135141802145600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280214s I0422 15:16:21.668831 135141802145600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.374 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28705835342407227 s) (per-host) I0422 15:16:21.668884 135141802145600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287123s (batch_requests_ready=0.002527s, total_serialization_initiated=0.284523s, others=0.000073s) I0422 15:16:21.669014 135141802145600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291114s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.291096) I0422 15:16:21.669833 135141802145600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.63 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/10. I0422 15:16:21.670177 135020115113728 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:36:21.670140 I0422 15:16:21.677415 135141802145600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:16:21.677699 135018512893696 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:16:21.677852 135141802145600 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776870981.0383267, 'wait_for_prev_duration_secs': 6.008148193359375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776870981.04035, 'checkpointer_blocking_duration_secs': 0.6299850940704346, 'get_old_steps_start_time': 1776870981.6703582, 'get_old_steps_duration_secs': 3.5762786865234375e-05, 'checkpoint_manager_blocking_start_time': 1776870981.0364616, 'checkpoint_manager_blocking_duration_secs': 0.6413555145263672} I0422 15:16:21.677964 135141802145600 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0422 15:16:21.678031 135141802145600 max_utils.py:750] Memstats: After params initialized: I0422 15:16:21.678080 135141802145600 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0422 15:16:21.678123 135141802145600 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0422 15:16:21.678150 135141802145600 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0422 15:16:21.678173 135141802145600 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0422 15:16:21.994859 135141802145600 metric_logger.py:196] completed step: 10, seconds: 11.126, TFLOP/s/device: 1.221, Tokens/s/device: 184.074, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0422 15:16:22.170786 135141802145600 metric_logger.py:196] completed step: 11, seconds: 1.110, TFLOP/s/device: 12.240, Tokens/s/device: 1845.010, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0422 15:16:22.318922 135141802145600 metric_logger.py:196] completed step: 12, seconds: 0.019, TFLOP/s/device: 719.848, Tokens/s/device: 108503.311, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0422 15:16:22.466470 135141802145600 metric_logger.py:196] completed step: 13, seconds: 0.165, TFLOP/s/device: 82.265, Tokens/s/device: 12399.947, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0422 15:16:22.472887 135141802145600 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0422 15:16:22.761326 135141802145600 checkpointing.py:798] Waited 0.2884104251861572 seconds for step 15 to finish before starting checkpointing. I0422 15:16:22.764016 135141802145600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:16:26.799565 135019032979200 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_6 I0422 15:17:03.973851 135020115113728 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.087 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.59204125404358 s) (per-host) I0422 15:17:03.973971 135020115113728 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.303675s. I0422 15:17:10.158343 135020115113728 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.488025s. I0422 15:17:10.158652 135018512893696 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:17:10.158771 135018512893696 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:17:10.158821 135018512893696 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:17:10.160470 135018512893696 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 15:17:10.160676 135141802145600 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0422 15:17:10.160871 135141802145600 checkpoint_manager.py:1452] Waiting for previous save to complete took 47.396852 seconds. If this number is high, consider checkpointing less frequently. I0422 15:17:10.162713 135141802145600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0422 15:17:10.165149 135141802145600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/15. I0422 15:17:10.467907 135141802145600 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:17:10.468001 135141802145600 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 I0422 15:17:10.509002 135141802145600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042683s I0422 15:17:10.509192 135141802145600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.139 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04953813552856445 s) (per-host) I0422 15:17:10.509243 135141802145600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049599s (batch_requests_ready=0.002585s, total_serialization_initiated=0.046946s, others=0.000068s) I0422 15:17:10.509336 135141802145600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054518s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=0.054498) I0422 15:17:10.510055 135141802145600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/15. I0422 15:17:10.510427 135018512893696 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:37:10.510386 I0422 15:17:10.516512 135141802145600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:17:10.516763 135013689448192 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:17:10.516888 135141802145600 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776870982.7639854, 'wait_for_prev_duration_secs': 47.39685249328613, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776871030.162752, 'checkpointer_blocking_duration_secs': 0.3478260040283203, 'get_old_steps_start_time': 1776871030.5105991, 'get_old_steps_duration_secs': 3.457069396972656e-05, 'checkpoint_manager_blocking_start_time': 1776870982.7617006, 'checkpoint_manager_blocking_duration_secs': 47.75515341758728} I0422 15:17:10.517002 135141802145600 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0422 15:17:10.518059 135141802145600 metric_logger.py:196] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.059, Tokens/s/device: 13876.185, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0422 15:17:10.525809 135141802145600 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.819, Tokens/s/device: 13839.895, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0422 15:17:16.015817 135019032979200 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_6 I0422 15:17:32.067004 135141802145600 metric_logger.py:196] completed step: 16, seconds: 48.051, TFLOP/s/device: 0.283, Tokens/s/device: 42.621, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0422 15:17:32.214765 135141802145600 metric_logger.py:196] completed step: 17, seconds: 21.397, TFLOP/s/device: 0.635, Tokens/s/device: 95.715, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0422 15:17:32.362336 135141802145600 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.144, Tokens/s/device: 13436.821, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0422 15:17:32.510575 135141802145600 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0422 15:17:32.511268 135141802145600 checkpointing.py:798] Waited 0.0007054805755615234 seconds for step 19 to finish before starting checkpointing. I0422 15:17:32.513324 135141802145600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:17:52.912683 135018512893696 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.208 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.452986001968384 s) (per-host) I0422 15:17:52.912805 135018512893696 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.402262s. I0422 15:17:58.673129 135018512893696 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.162550s. I0422 15:17:58.673371 135013689448192 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:17:58.673435 135013689448192 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:17:58.673482 135013689448192 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:17:58.675725 135013689448192 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 15:17:58.675895 135141802145600 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0422 15:17:58.676021 135141802145600 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.162701 seconds. If this number is high, consider checkpointing less frequently. I0422 15:17:58.677411 135141802145600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0422 15:17:58.679492 135141802145600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/19. I0422 15:17:58.967991 135141802145600 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:17:58.968085 135141802145600 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 I0422 15:17:59.009399 135141802145600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042969s I0422 15:17:59.009566 135141802145600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 30.906 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.049912214279174805 s) (per-host) I0422 15:17:59.009617 135141802145600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049974s (batch_requests_ready=0.002686s, total_serialization_initiated=0.047221s, others=0.000067s) I0422 15:17:59.009714 135141802145600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054094s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.054083) I0422 15:17:59.010369 135141802145600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.33 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/19. I0422 15:17:59.010704 135019032979200 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:37:59.010667 I0422 15:17:59.012599 135141802145600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:17:59.012864 135013689448192 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:17:59.012998 135141802145600 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776871052.513295, 'wait_for_prev_duration_secs': 26.16270136833191, 'time_between_consecutive_saves_sec': 22.35278582572937, 'checkpointer_blocking_start_time': 1776871078.6774507, 'checkpointer_blocking_duration_secs': 0.33339476585388184, 'get_old_steps_start_time': 1776871079.010862, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776871052.511515, 'checkpoint_manager_blocking_duration_secs': 26.50145435333252} I0422 15:17:59.013174 135141802145600 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0422 15:17:59.013218 135141802145600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:18:04.189681 135013111346944 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_6 I0422 15:18:41.379476 135019032979200 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.237 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.41978311538696 s) (per-host) I0422 15:18:41.379614 135019032979200 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.368799s. I0422 15:18:47.712019 135019032979200 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.701187s. I0422 15:18:47.712282 135013689448192 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:18:47.712350 135013689448192 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:18:47.712393 135013689448192 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:18:47.713996 135013689448192 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 15:18:47.714157 135141802145600 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0422 15:18:47.714328 135141802145600 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:18:47.715267 135141802145600 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.797, Tokens/s/device: 13836.622, total_weights: 65536, loss: 5.385, lm_loss: 5.385, perplexity: 218.045 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Wed Apr 22 15:19:00 UTC 2026 EXIT_CODE=0