XPK Start: Wed Apr 22 15:41:55 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 15:42:19.880977: 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 15:42:20.422830 134931968153408 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 15:42:29.464398 134931968153408 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 15:42:29.466819 134931968153408 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-2bsnp-slice-job-0-0.mt-16-zarr3-false-rs-2bsnp:8482 I0422 15:42:30.933575 134931968153408 max_utils.py:284] Jax distributed system initialized! I0422 15:42:36.972068 134931968153408 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 15:42:36.972301 134931968153408 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 15:42:36.972343 134931968153408 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 15:42:36.972377 134931968153408 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 15:42:37.671147 134931968153408 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 15:42:37.785387 134931968153408 checkpointing.py:688] Setting up checkpoint logger... I0422 15:42:37.785502 134931968153408 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0422 15:42:37.785550 134931968153408 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 15:42:37.785753 134931968153408 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7ab7ae11c800>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 15:42:39.100823 134931968153408 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 15:42:39.101058 134931968153408 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 0x7aa28433bb60>}, handler_registry=None I0422 15:42:39.101313 134931968153408 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa28433bb60>` 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 15:42:39.101361 134931968153408 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa2842f6000>` 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 15:42:39.101398 134931968153408 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 0x7aa28433bb60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa28433bb60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa2842f6000>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa2842f6000>}). I0422 15:42:39.101731 134931968153408 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 15:42:39.101802 134931968153408 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7aa2842b59e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 15:42:41.093232 134810668349184 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776872395041488978, 'commit_timestamp_nsecs': 1776872441194350389, '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_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0422 15:42:41.350476 134810668349184 checkpoint_manager.py:1871] Missing metrics for step 9 I0422 15:42:41.752429 134931968153408 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_16_zarr3_false/checkpoints I0422 15:42:42.221395 134931968153408 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_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7aa27c5a8500> I0422 15:42:42.221574 134931968153408 checkpointing.py:302] Checkpoint manager created! I0422 15:42:43.183417 134931968153408 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 15:42:43.183532 134931968153408 checkpointing.py:582] restoring from this run's directory step 9 I0422 15:42:43.691871 134931968153408 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_16_zarr3_false/checkpoints/9. I0422 15:42:43.954993 134931968153408 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_16_zarr3_false/checkpoints/9. I0422 15:42:43.955435 134931968153408 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.26 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_16_zarr3_false/checkpoints/9. I0422 15:42:45.353672 1447 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 15:42:51.999452 134931968153408 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 237.550 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.6524059772491455 s) (per-host) I0422 15:42:52.006939 134931968153408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 224.022 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.06450343132019 s) (per-host) I0422 15:42:54.322518 134931968153408 event_tracking.py:138] [process=6] [sync] Finished load in 10.63 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_false/checkpoints/9 I0422 15:42:54.322674 134931968153408 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_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776872563.6917741, 'checkpointer_duration_secs': 10.630866765975952, 'checkpoint_manager_start_time': 1776872563.1841495, 'checkpoint_manager_duration_secs': 11.138493776321411} fsdp: 32 I0422 15:42:54.398245 134931968153408 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0422 15:42:54.398363 134931968153408 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:42:54.398413 134931968153408 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:42:54.398456 134931968153408 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0422 15:42:54.398491 134931968153408 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:42:54.398523 134931968153408 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:42:54.398557 134931968153408 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:42:54.398591 134931968153408 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0422 15:42:54.398624 134931968153408 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:42:54.398656 134931968153408 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:42:54.398688 134931968153408 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0422 15:42:54.398720 134931968153408 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0422 15:42:54.796875 134931968153408 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:42:54.796966 134931968153408 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:42:54.802432 134931968153408 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:42:54.802488 134931968153408 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:42:54.818817 134931968153408 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:42:54.818875 134931968153408 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:42:54.834567 134931968153408 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:42:54.834622 134931968153408 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:42:54.858290 134931968153408 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 15:42:54.858364 134931968153408 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:42:54.874395 134931968153408 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 15:42:54.874460 134931968153408 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:42:54.890129 134931968153408 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 15:42:54.890186 134931968153408 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:42:54.920570 134931968153408 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 15:42:54.920643 134931968153408 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:42:54.940548 134931968153408 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 15:42:54.940622 134931968153408 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 15:43:06.569616 134931968153408 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 15:43:06.571973 134931968153408 metric_logger.py:301] number parameters: 1.104 billion I0422 15:43:17.753617 134931968153408 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0422 15:43:17.905983 134931968153408 checkpointing.py:798] Waited 0.15235495567321777 seconds for step 10 to finish before starting checkpointing. I0422 15:43:17.908585 134931968153408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:43:17.910801 134931968153408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0422 15:43:17.912579 134931968153408 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_16_zarr3_false/checkpoints/10. I0422 15:43:18.381146 134931968153408 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 15:43:18.382174 134931968153408 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:43:18.382240 134931968153408 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 15:43:18.663084 134931968153408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282718s I0422 15:43:18.663303 134931968153408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.322 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2898247241973877 s) (per-host) I0422 15:43:18.663374 134931968153408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.289910s (batch_requests_ready=0.002596s, total_serialization_initiated=0.287219s, others=0.000095s) I0422 15:43:18.663500 134931968153408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294196s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.294178) I0422 15:43:18.664250 134931968153408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 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_16_zarr3_false/checkpoints/10. I0422 15:43:18.664619 134805870081792 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 16:03:18.664578 I0422 15:43:18.688470 134931968153408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:43:18.688743 134809577821952 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:43:18.688913 134931968153408 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776872597.9085655, 'wait_for_prev_duration_secs': 6.246566772460938e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776872597.9108849, 'checkpointer_blocking_duration_secs': 0.7538955211639404, 'get_old_steps_start_time': 1776872598.664804, 'get_old_steps_duration_secs': 3.504753112792969e-05, 'checkpoint_manager_blocking_start_time': 1776872597.9063756, 'checkpoint_manager_blocking_duration_secs': 0.7825019359588623} I0422 15:43:18.689021 134931968153408 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0422 15:43:18.689110 134931968153408 max_utils.py:750] Memstats: After params initialized: I0422 15:43:18.689162 134931968153408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0422 15:43:18.689194 134931968153408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0422 15:43:18.689220 134931968153408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0422 15:43:18.689244 134931968153408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0422 15:43:19.010408 134931968153408 metric_logger.py:196] completed step: 10, seconds: 11.181, TFLOP/s/device: 1.215, Tokens/s/device: 183.175, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0422 15:43:19.198800 134931968153408 metric_logger.py:196] completed step: 11, seconds: 1.255, TFLOP/s/device: 10.828, Tokens/s/device: 1632.118, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0422 15:43:19.346361 134931968153408 metric_logger.py:196] completed step: 12, seconds: 0.031, TFLOP/s/device: 431.872, Tokens/s/device: 65096.469, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0422 15:43:19.494048 134931968153408 metric_logger.py:196] completed step: 13, seconds: 0.165, TFLOP/s/device: 82.277, Tokens/s/device: 12401.749, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0422 15:43:19.500442 134931968153408 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0422 15:43:19.789052 134931968153408 checkpointing.py:798] Waited 0.28858280181884766 seconds for step 15 to finish before starting checkpointing. I0422 15:43:19.791754 134931968153408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:43:24.104387 134806356309760 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_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6 I0422 15:44:00.562243 134805870081792 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.441 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.18873429298401 s) (per-host) I0422 15:44:00.562369 134805870081792 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.897632s. I0422 15:44:10.282928 134805870081792 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 51.618176s. I0422 15:44:10.283180 134809577821952 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:44:10.283236 134809577821952 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:44:10.283283 134809577821952 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:44:10.285007 134809577821952 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 15:44:10.285206 134931968153408 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 15:44:10.285351 134931968153408 checkpoint_manager.py:1452] Waiting for previous save to complete took 50.493598 seconds. If this number is high, consider checkpointing less frequently. I0422 15:44:10.287463 134931968153408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0422 15:44:10.290018 134931968153408 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_16_zarr3_false/checkpoints/15. I0422 15:44:11.008749 134931968153408 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:44:11.008845 134931968153408 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 15:44:11.036927 134931968153408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029844s I0422 15:44:11.037054 134931968153408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 42.203 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03655099868774414 s) (per-host) I0422 15:44:11.037116 134931968153408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.036606s (batch_requests_ready=0.002568s, total_serialization_initiated=0.033977s, others=0.000062s) I0422 15:44:11.037218 134931968153408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041313s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=0.041299) I0422 15:44:11.037968 134931968153408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 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_16_zarr3_false/checkpoints/15. I0422 15:44:11.038300 134809577821952 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 16:04:11.038265 I0422 15:44:11.061754 134931968153408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:44:11.062052 134806356309760 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:44:11.062231 134931968153408 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776872599.7917237, 'wait_for_prev_duration_secs': 50.49359750747681, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776872650.287502, 'checkpointer_blocking_duration_secs': 0.7509410381317139, 'get_old_steps_start_time': 1776872651.0384672, 'get_old_steps_duration_secs': 3.6716461181640625e-05, 'checkpoint_manager_blocking_start_time': 1776872599.7894456, 'checkpoint_manager_blocking_duration_secs': 51.27274966239929} I0422 15:44:11.062346 134931968153408 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0422 15:44:11.063529 134931968153408 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.259, Tokens/s/device: 13906.242, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0422 15:44:11.071810 134931968153408 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.894, Tokens/s/device: 13851.315, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0422 15:44:16.488531 134804242683648 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_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6 I0422 15:44:33.016759 134931968153408 metric_logger.py:196] completed step: 16, seconds: 51.570, TFLOP/s/device: 0.263, Tokens/s/device: 39.713, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0422 15:44:33.164492 134931968153408 metric_logger.py:196] completed step: 17, seconds: 21.800, TFLOP/s/device: 0.623, Tokens/s/device: 93.945, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0422 15:44:33.311937 134931968153408 metric_logger.py:196] completed step: 18, seconds: 0.154, TFLOP/s/device: 88.486, Tokens/s/device: 13337.588, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0422 15:44:33.460021 134931968153408 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0422 15:44:33.460730 134931968153408 checkpointing.py:798] Waited 0.0007202625274658203 seconds for step 19 to finish before starting checkpointing. I0422 15:44:33.462717 134931968153408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:44:53.683737 134809577821952 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.007 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.683194398880005 s) (per-host) I0422 15:44:53.683868 134809577821952 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.645457s. I0422 15:45:01.637327 134809577821952 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.598899s. I0422 15:45:01.637643 134806356309760 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:45:01.637758 134806356309760 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:45:01.637806 134806356309760 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:45:01.639292 134806356309760 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 15:45:01.639457 134931968153408 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 15:45:01.639578 134931968153408 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.176868 seconds. If this number is high, consider checkpointing less frequently. I0422 15:45:01.641066 134931968153408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0422 15:45:01.643181 134931968153408 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_16_zarr3_false/checkpoints/19. I0422 15:45:01.956773 134931968153408 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:45:01.956866 134931968153408 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 15:45:01.982430 134931968153408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027234s I0422 15:45:01.982562 134931968153408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 45.167 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03415274620056152 s) (per-host) I0422 15:45:01.982608 134931968153408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.034209s (batch_requests_ready=0.002694s, total_serialization_initiated=0.031455s, others=0.000060s) I0422 15:45:01.982686 134931968153408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038259s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.038247) I0422 15:45:01.983392 134931968153408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.34 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_16_zarr3_false/checkpoints/19. I0422 15:45:01.983726 134806356309760 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 16:05:01.983689 I0422 15:45:01.989334 134931968153408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:45:01.989623 134804242683648 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:45:01.989778 134931968153408 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776872673.4626868, 'wait_for_prev_duration_secs': 28.176868438720703, 'time_between_consecutive_saves_sec': 23.17763614654541, 'checkpointer_blocking_start_time': 1776872701.6411166, 'checkpointer_blocking_duration_secs': 0.3427600860595703, 'get_old_steps_start_time': 1776872701.9838946, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776872673.4609745, 'checkpoint_manager_blocking_duration_secs': 28.52877116203308} I0422 15:45:01.989899 134931968153408 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0422 15:45:01.989941 134931968153408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:45:07.653550 134804259469056 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_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6 I0422 15:45:44.252347 134806356309760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.339 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.3038969039917 s) (per-host) I0422 15:45:44.252477 134806356309760 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.268639s. I0422 15:45:53.269139 134806356309760 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 51.285282s. I0422 15:45:53.269424 134804242683648 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:45:53.269539 134804242683648 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:45:53.269581 134804242683648 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:45:53.270982 134804242683648 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 15:45:53.271141 134931968153408 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 15:45:53.271303 134931968153408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:45:53.272216 134931968153408 metric_logger.py:196] completed step: 19, seconds: 0.149, TFLOP/s/device: 91.338, Tokens/s/device: 13767.419, 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 15:46:05 UTC 2026 EXIT_CODE=0