XPK Start: Thu Apr 23 15:21:03 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. `rope_parameters`'s factor field must be a float >= 1, got 40 `rope_parameters`'s beta_fast field must be a float, got 32 `rope_parameters`'s beta_slow field must be a float, got 1 DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 2026-04-23 15:21:28.459712: 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) I0423 15:21:28.669544 134908560410432 max_utils.py:273] Attempting to initialize the jax distributed system... I0423 15:21:37.709928 134908560410432 distributed.py:149] Starting JAX distributed service on [::]:8482 I0423 15:21:37.712262 134908560410432 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-fvyme-slice-job-0-0.mt-15-ocdbt-true-rs-fvyme:8482 I0423 15:21:38.663643 134908560410432 max_utils.py:284] Jax distributed system initialized! I0423 15:21:44.827128 134908560410432 max_utils.py:800] System Information: Jax Version: 0.9.2 I0423 15:21:44.827236 134908560410432 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0423 15:21:44.827277 134908560410432 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 I0423 15:21:44.827313 134908560410432 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 15:21:45.519376 134908560410432 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 15:21:45.631058 134908560410432 checkpointing.py:688] Setting up checkpoint logger... I0423 15:21:45.631193 134908560410432 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 15:21:45.631238 134908560410432 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0423 15:21:45.631446 134908560410432 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 0x7ab237d18620>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 15:21:46.954823 134908560410432 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 15:21:46.955057 134908560410432 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 0x7a9de0164e90>}, handler_registry=None I0423 15:21:46.955315 134908560410432 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a9de0164e90>` 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`. I0423 15:21:46.955365 134908560410432 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9cf839c440>` 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`. I0423 15:21:46.955403 134908560410432 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 0x7a9de0164e90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a9de0164e90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9cf839c440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9cf839c440>}). I0423 15:21:46.955746 134908560410432 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0423 15:21:46.955819 134908560410432 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a9de014f560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0423 15:21:48.578655 134786987312896 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776957543734163148, 'commit_timestamp_nsecs': 1776957590551501651, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0423 15:21:48.836154 134786987312896 checkpoint_manager.py:1871] Missing metrics for step 9 I0423 15:21:49.272008 134908560410432 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints I0423 15:21:49.659526 134908560410432 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a9de01663c0> I0423 15:21:49.659708 134908560410432 checkpointing.py:302] Checkpoint manager created! I0423 15:21:50.630271 134908560410432 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 15:21:50.630395 134908560410432 checkpointing.py:582] restoring from this run's directory step 9 I0423 15:21:50.900161 134908560410432 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/9. I0423 15:21:51.171150 134908560410432 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/9. I0423 15:21:51.171568 134908560410432 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.27 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/9. I0423 15:21:52.623950 1457 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 15:21:59.990892 134908560410432 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 214.318 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.373523473739624 s) (per-host) I0423 15:21:59.999391 134908560410432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 202.875 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.800871849060059 s) (per-host) I0423 15:22:01.098911 134908560410432 event_tracking.py:138] [process=6] [sync] Finished load in 10.20 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/9 I0423 15:22:01.099068 134908560410432 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776957710.9000428, 'checkpointer_duration_secs': 10.198994398117065, 'checkpoint_manager_start_time': 1776957710.6310318, 'checkpoint_manager_duration_secs': 10.468007802963257} fsdp: 32 I0423 15:22:01.174792 134908560410432 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 15:22:01.174906 134908560410432 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 15:22:01.174957 134908560410432 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 15:22:01.174998 134908560410432 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 15:22:01.175035 134908560410432 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 15:22:01.175070 134908560410432 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 15:22:01.175117 134908560410432 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:22:01.175152 134908560410432 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 15:22:01.175188 134908560410432 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:22:01.175221 134908560410432 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:22:01.175255 134908560410432 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 15:22:01.175295 134908560410432 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 15:22:01.585168 134908560410432 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 15:22:01.585268 134908560410432 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:22:01.590635 134908560410432 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 15:22:01.590692 134908560410432 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:22:01.607115 134908560410432 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 15:22:01.607174 134908560410432 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:22:01.622904 134908560410432 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 15:22:01.622965 134908560410432 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:22:01.647782 134908560410432 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:22:01.647850 134908560410432 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:22:01.663678 134908560410432 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:22:01.663737 134908560410432 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:22:01.679536 134908560410432 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:22:01.679594 134908560410432 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:22:01.708990 134908560410432 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 15:22:01.709063 134908560410432 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:22:01.734116 134908560410432 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 15:22:01.734217 134908560410432 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 15:22:13.372547 134908560410432 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. I0423 15:22:13.374799 134908560410432 metric_logger.py:301] number parameters: 1.104 billion I0423 15:22:24.981386 134908560410432 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0423 15:22:25.133533 134908560410432 checkpointing.py:798] Waited 0.1521298885345459 seconds for step 10 to finish before starting checkpointing. I0423 15:22:25.135743 134908560410432 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 15:22:25.137747 134908560410432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0423 15:22:25.139115 134908560410432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/10. I0423 15:22:25.475166 134908560410432 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 15:22:25.476115 134908560410432 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:22:25.476173 134908560410432 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 I0423 15:22:25.756960 134908560410432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282560s I0423 15:22:25.757153 134908560410432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.330 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28943514823913574 s) (per-host) I0423 15:22:25.757211 134908560410432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.289506s (batch_requests_ready=0.002566s, total_serialization_initiated=0.286858s, others=0.000083s) I0423 15:22:25.757313 134908560410432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.293453s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.293438) I0423 15:22:25.758061 134908560410432 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/10. I0423 15:22:25.758424 134782619858688 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 15:42:25.758386 I0423 15:22:25.766901 134908560410432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:22:25.767163 134782113543936 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:22:25.767314 134908560410432 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776957745.1357255, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776957745.1377854, 'checkpointer_blocking_duration_secs': 0.6208071708679199, 'get_old_steps_start_time': 1776957745.7586186, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776957745.133914, 'checkpoint_manager_blocking_duration_secs': 0.6333658695220947} I0423 15:22:25.767424 134908560410432 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0423 15:22:25.767491 134908560410432 max_utils.py:750] Memstats: After params initialized: I0423 15:22:25.767539 134908560410432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 15:22:25.767570 134908560410432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 15:22:25.767602 134908560410432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 15:22:25.767626 134908560410432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 15:22:26.087895 134908560410432 metric_logger.py:196] completed step: 10, seconds: 11.606, TFLOP/s/device: 1.171, Tokens/s/device: 176.468, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0423 15:22:26.262782 134908560410432 metric_logger.py:196] completed step: 11, seconds: 1.105, TFLOP/s/device: 12.298, Tokens/s/device: 1853.748, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0423 15:22:26.410774 134908560410432 metric_logger.py:196] completed step: 12, seconds: 0.017, TFLOP/s/device: 781.184, Tokens/s/device: 117748.520, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0423 15:22:26.558146 134908560410432 metric_logger.py:196] completed step: 13, seconds: 0.166, TFLOP/s/device: 81.789, Tokens/s/device: 12328.066, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0423 15:22:26.565144 134908560410432 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0423 15:22:26.853749 134908560410432 checkpointing.py:798] Waited 0.2885763645172119 seconds for step 15 to finish before starting checkpointing. I0423 15:22:26.856493 134908560410432 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:22:30.907721 134782603073280 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_6 I0423 15:23:07.273076 134782619858688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.785 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.80532240867615 s) (per-host) I0423 15:23:07.273219 134782619858688 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.514679s. I0423 15:23:15.334680 134782619858688 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.576124s. I0423 15:23:15.334963 134782113543936 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:23:15.335089 134782113543936 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:23:15.335176 134782113543936 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:23:15.337757 134782113543936 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 15:23:15.337944 134908560410432 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. W0423 15:23:15.338076 134908560410432 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.481582 seconds. If this number is high, consider checkpointing less frequently. I0423 15:23:15.340162 134908560410432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0423 15:23:15.342592 134908560410432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/15. I0423 15:23:16.059036 134908560410432 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:23:16.059147 134908560410432 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 I0423 15:23:16.086440 134908560410432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029005s I0423 15:23:16.086575 134908560410432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 43.168 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.035733938217163086 s) (per-host) I0423 15:23:16.086641 134908560410432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035809s (batch_requests_ready=0.002587s, total_serialization_initiated=0.033140s, others=0.000082s) I0423 15:23:16.086733 134908560410432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040922s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.040909) I0423 15:23:16.087463 134908560410432 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/15. I0423 15:23:16.087832 134782113543936 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 15:43:16.087786 I0423 15:23:16.108648 134908560410432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:23:16.108913 134782603073280 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:23:16.109073 134908560410432 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776957746.8564625, 'wait_for_prev_duration_secs': 48.481582164764404, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776957795.3402486, 'checkpointer_blocking_duration_secs': 0.7477326393127441, 'get_old_steps_start_time': 1776957796.0880065, 'get_old_steps_duration_secs': 3.3855438232421875e-05, 'checkpoint_manager_blocking_start_time': 1776957746.854119, 'checkpoint_manager_blocking_duration_secs': 49.25491905212402} I0423 15:23:16.109206 134908560410432 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0423 15:23:16.110313 134908560410432 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.619, Tokens/s/device: 13960.559, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0423 15:23:16.118395 134908560410432 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.499, Tokens/s/device: 13791.710, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0423 15:23:21.217593 134778355447552 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_6 I0423 15:23:37.749300 134908560410432 metric_logger.py:196] completed step: 16, seconds: 49.551, TFLOP/s/device: 0.274, Tokens/s/device: 41.331, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0423 15:23:37.896996 134908560410432 metric_logger.py:196] completed step: 17, seconds: 21.486, TFLOP/s/device: 0.632, Tokens/s/device: 95.317, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0423 15:23:38.044371 134908560410432 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.910, Tokens/s/device: 13401.475, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0423 15:23:38.192579 134908560410432 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0423 15:23:38.193302 134908560410432 checkpointing.py:798] Waited 0.0007340908050537109 seconds for step 19 to finish before starting checkpointing. I0423 15:23:38.195745 134908560410432 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:23:57.599933 134782113543936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.018 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.549063205718994 s) (per-host) I0423 15:23:57.600047 134782113543936 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.512099s. I0423 15:24:06.580947 134782113543936 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.492982s. I0423 15:24:06.581279 134782603073280 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:24:06.581397 134782603073280 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:24:06.581446 134782603073280 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:24:06.582551 134782603073280 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0423 15:24:06.582707 134908560410432 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. W0423 15:24:06.582828 134908560410432 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.387091 seconds. If this number is high, consider checkpointing less frequently. I0423 15:24:06.584616 134908560410432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0423 15:24:06.586730 134908560410432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/19. I0423 15:24:06.889206 134908560410432 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:24:06.889296 134908560410432 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 I0423 15:24:06.915194 134908560410432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027513s I0423 15:24:06.915319 134908560410432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 44.866 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03438162803649902 s) (per-host) I0423 15:24:06.915372 134908560410432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.034444s (batch_requests_ready=0.002696s, total_serialization_initiated=0.031682s, others=0.000066s) I0423 15:24:06.915450 134908560410432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038654s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.038642) I0423 15:24:06.916134 134908560410432 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/19. I0423 15:24:06.916474 134782603073280 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 15:44:06.916437 I0423 15:24:06.922635 134908560410432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:24:06.922952 134778355447552 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:24:06.923143 134908560410432 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776957818.1957161, 'wait_for_prev_duration_secs': 28.3870906829834, 'time_between_consecutive_saves_sec': 22.857916116714478, 'checkpointer_blocking_start_time': 1776957846.584657, 'checkpointer_blocking_duration_secs': 0.3319580554962158, 'get_old_steps_start_time': 1776957846.9166334, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776957818.1935515, 'checkpoint_manager_blocking_duration_secs': 28.729557275772095} I0423 15:24:06.923269 134908560410432 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0423 15:24:06.923316 134908560410432 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:24:12.492979 134782079973120 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_6 I0423 15:24:49.846747 134782603073280 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.764 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.96576547622681 s) (per-host) I0423 15:24:49.846869 134782603073280 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.930284s. I0423 15:24:56.049725 134782603073280 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.133122s. I0423 15:24:56.050010 134778355447552 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:24:56.050141 134778355447552 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:24:56.050186 134778355447552 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:24:56.052763 134778355447552 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 15:24:56.052934 134908560410432 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. I0423 15:24:56.053128 134908560410432 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 15:24:56.054077 134908560410432 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.886, Tokens/s/device: 13850.003, 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: Thu Apr 23 15:25:08 UTC 2026 EXIT_CODE=0