XPK Start: Fri Apr 24 14:53:04 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. 2026-04-24 14:53:29.622742: 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) I0424 14:53:29.833854 133126962399040 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:53:38.877047 133126962399040 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:53:38.879448 133126962399040 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-ipoqu-slice-job-0-0.mt-16-zarr3-false-rs-ipoqu:8482 I0424 14:53:39.683892 133126962399040 max_utils.py:284] Jax distributed system initialized! I0424 14:53:45.935680 133126962399040 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:53:45.935787 133126962399040 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:53:45.935829 133126962399040 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 I0424 14:53:45.935865 133126962399040 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:53:46.635611 133126962399040 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:53:46.748243 133126962399040 checkpointing.py:688] Setting up checkpoint logger... I0424 14:53:46.748363 133126962399040 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0424 14:53:46.748408 133126962399040 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:53:46.748610 133126962399040 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 0x791366518e30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:53:48.005351 133126962399040 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:53:48.005583 133126962399040 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 0x78fdf85e12e0>}, handler_registry=None I0424 14:53:48.005824 133126962399040 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78fdf85e12e0>` 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`. I0424 14:53:48.005872 133126962399040 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ff00704200>` 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`. I0424 14:53:48.005907 133126962399040 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 0x78fdf85e12e0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78fdf85e12e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ff00704200>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ff00704200>}). I0424 14:53:48.006250 133126962399040 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:53:48.006330 133126962399040 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78ff3012f560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:53:49.606449 133005196650240 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777042262411534816, 'commit_timestamp_nsecs': 1777042308015540027, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0424 14:53:49.874501 133005557339904 checkpoint_manager.py:1871] Missing metrics for step 0 I0424 14:53:50.158926 133126962399040 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints I0424 14:53:50.289180 133126962399040 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78ff30072090> I0424 14:53:50.289364 133126962399040 checkpointing.py:302] Checkpoint manager created! I0424 14:53:51.256489 133126962399040 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:53:51.256615 133126962399040 checkpointing.py:582] restoring from this run's directory step 9 I0424 14:53:51.483606 133126962399040 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 14:53:51.733747 133126962399040 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 14:53:51.734188 133126962399040 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 14:53:53.219192 1456 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:54:00.209460 133126962399040 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 225.842 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.997253894805908 s) (per-host) I0424 14:54:00.217647 133126962399040 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 211.322 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.48906397819519 s) (per-host) I0424 14:54:00.599380 133126962399040 event_tracking.py:138] [process=6] [sync] Finished load in 9.12 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9 I0424 14:54:00.599543 133126962399040 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1777042431.4835, 'checkpointer_duration_secs': 9.11600947380066, 'checkpoint_manager_start_time': 1777042431.257226, 'checkpoint_manager_duration_secs': 9.342285871505737} fsdp: 32 I0424 14:54:00.674992 133126962399040 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:54:00.675119 133126962399040 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:54:00.675172 133126962399040 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:54:00.675215 133126962399040 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:54:00.675257 133126962399040 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:54:00.675292 133126962399040 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:54:00.675328 133126962399040 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:54:00.675364 133126962399040 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:54:00.675399 133126962399040 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:54:00.675432 133126962399040 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:54:00.675466 133126962399040 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:54:00.675498 133126962399040 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:54:01.081161 133126962399040 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:54:01.081261 133126962399040 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:54:01.086718 133126962399040 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:54:01.086775 133126962399040 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:54:01.103452 133126962399040 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:54:01.103510 133126962399040 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:54:01.119768 133126962399040 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:54:01.119830 133126962399040 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:54:01.145234 133126962399040 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:54:01.145301 133126962399040 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:54:01.161622 133126962399040 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:54:01.161684 133126962399040 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:54:01.177743 133126962399040 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:54:01.177802 133126962399040 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:54:01.207934 133126962399040 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:54:01.208006 133126962399040 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:54:01.228109 133126962399040 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:54:01.228173 133126962399040 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:54:12.839035 133126962399040 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. I0424 14:54:12.841372 133126962399040 metric_logger.py:301] number parameters: 1.104 billion I0424 14:54:24.466862 133126962399040 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 14:54:24.618515 133126962399040 checkpointing.py:798] Waited 0.1516268253326416 seconds for step 10 to finish before starting checkpointing. I0424 14:54:24.621030 133126962399040 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:54:24.623005 133126962399040 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0424 14:54:24.624443 133126962399040 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10. I0424 14:54:24.958266 133126962399040 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:54:24.959316 133126962399040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:54:24.959385 133126962399040 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 I0424 14:54:25.235673 133126962399040 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278195s I0424 14:54:25.235850 133126962399040 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.409 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28519558906555176 s) (per-host) I0424 14:54:25.235903 133126962399040 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285261s (batch_requests_ready=0.002718s, total_serialization_initiated=0.282469s, others=0.000073s) I0424 14:54:25.236003 133126962399040 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289391s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.289374) I0424 14:54:25.236758 133126962399040 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10. I0424 14:54:25.237050 133001379837696 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:14:25.237015 I0424 14:54:25.242356 133126962399040 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:54:25.242647 133005163079424 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:54:25.242797 133126962399040 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042464.621012, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777042464.6230438, 'checkpointer_blocking_duration_secs': 0.6141767501831055, 'get_old_steps_start_time': 1777042465.2372453, 'get_old_steps_duration_secs': 3.2901763916015625e-05, 'checkpoint_manager_blocking_start_time': 1777042464.6189167, 'checkpoint_manager_blocking_duration_secs': 0.6238448619842529} I0424 14:54:25.242903 133126962399040 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 14:54:25.242971 133126962399040 max_utils.py:750] Memstats: After params initialized: I0424 14:54:25.243021 133126962399040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 14:54:25.243052 133126962399040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 14:54:25.243078 133126962399040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 14:54:25.243117 133126962399040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 14:54:25.563296 133126962399040 metric_logger.py:196] completed step: 10, seconds: 11.624, TFLOP/s/device: 1.169, Tokens/s/device: 176.181, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0424 14:54:25.733996 133126962399040 metric_logger.py:196] completed step: 11, seconds: 1.095, TFLOP/s/device: 12.413, Tokens/s/device: 1870.981, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0424 14:54:25.881481 133126962399040 metric_logger.py:196] completed step: 12, seconds: 0.014, TFLOP/s/device: 989.811, Tokens/s/device: 149195.017, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0424 14:54:26.028973 133126962399040 metric_logger.py:196] completed step: 13, seconds: 0.166, TFLOP/s/device: 81.799, Tokens/s/device: 12329.699, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0424 14:54:26.035365 133126962399040 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 14:54:26.324495 133126962399040 checkpointing.py:798] Waited 0.28910112380981445 seconds for step 15 to finish before starting checkpointing. I0424 14:54:26.327216 133126962399040 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:54:30.179219 133004013836032 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6 I0424 14:55:07.227536 133001379837696 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.363 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.276829957962036 s) (per-host) I0424 14:55:07.227663 133001379837696 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.990480s. I0424 14:55:16.295855 133001379837696 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 51.058654s. I0424 14:55:16.296163 133005163079424 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:55:16.296299 133005163079424 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:55:16.296348 133005163079424 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:55:16.298730 133005163079424 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 14:55:16.298933 133126962399040 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. W0424 14:55:16.299065 133126962399040 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.971849 seconds. If this number is high, consider checkpointing less frequently. I0424 14:55:16.300971 133126962399040 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0424 14:55:16.303340 133126962399040 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15. I0424 14:55:16.624835 133126962399040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:55:16.624928 133126962399040 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 I0424 14:55:16.651555 133126962399040 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028278s I0424 14:55:16.651724 133126962399040 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 43.823 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03520011901855469 s) (per-host) I0424 14:55:16.651776 133126962399040 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035261s (batch_requests_ready=0.002755s, total_serialization_initiated=0.032437s, others=0.000069s) I0424 14:55:16.651871 133126962399040 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040149s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.040136) I0424 14:55:16.652457 133126962399040 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15. I0424 14:55:16.652801 133004013836032 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:15:16.652762 I0424 14:55:16.675857 133126962399040 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:55:16.676135 133005163079424 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:55:16.676270 133126962399040 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042466.327187, 'wait_for_prev_duration_secs': 49.97184872627258, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777042516.3010125, 'checkpointer_blocking_duration_secs': 0.351942777633667, 'get_old_steps_start_time': 1777042516.652975, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1777042466.3248558, 'checkpoint_manager_blocking_duration_secs': 50.35138177871704} I0424 14:55:16.676377 133126962399040 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 14:55:16.677358 133126962399040 metric_logger.py:196] completed step: 14, seconds: 0.146, TFLOP/s/device: 93.068, Tokens/s/device: 14028.166, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0424 14:55:16.684844 133126962399040 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.936, Tokens/s/device: 13857.594, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0424 14:55:21.921868 132998687090432 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6 I0424 14:55:38.422699 133126962399040 metric_logger.py:196] completed step: 16, seconds: 50.648, TFLOP/s/device: 0.268, Tokens/s/device: 40.436, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0424 14:55:38.570145 133126962399040 metric_logger.py:196] completed step: 17, seconds: 21.592, TFLOP/s/device: 0.629, Tokens/s/device: 94.850, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0424 14:55:38.717645 133126962399040 metric_logger.py:196] completed step: 18, seconds: 0.154, TFLOP/s/device: 88.462, Tokens/s/device: 13333.941, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0424 14:55:38.865833 133126962399040 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 14:55:38.866520 133126962399040 checkpointing.py:798] Waited 0.0006978511810302734 seconds for step 19 to finish before starting checkpointing. I0424 14:55:38.868524 133126962399040 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:55:58.376446 133004013836032 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.826 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.759881019592285 s) (per-host) I0424 14:55:58.376567 133004013836032 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.723647s. I0424 14:56:07.040067 133004013836032 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.387131s. I0424 14:56:07.040294 133005163079424 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:56:07.040353 133005163079424 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:56:07.040416 133005163079424 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:56:07.042229 133005163079424 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 14:56:07.042398 133126962399040 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. W0424 14:56:07.042529 133126962399040 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.174010 seconds. If this number is high, consider checkpointing less frequently. I0424 14:56:07.044136 133126962399040 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0424 14:56:07.046084 133126962399040 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19. I0424 14:56:07.358630 133126962399040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:56:07.358721 133126962399040 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 I0424 14:56:07.383961 133126962399040 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026975s I0424 14:56:07.384086 133126962399040 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 45.356 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03401041030883789 s) (per-host) I0424 14:56:07.384148 133126962399040 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.034080s (batch_requests_ready=0.002909s, total_serialization_initiated=0.031097s, others=0.000074s) I0424 14:56:07.384226 133126962399040 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038377s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.038366) I0424 14:56:07.384998 133126962399040 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19. I0424 14:56:07.385278 133005163079424 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:16:07.385255 I0424 14:56:07.392087 133126962399040 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:56:07.392387 132998687090432 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:56:07.392544 133126962399040 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042538.8684947, 'wait_for_prev_duration_secs': 28.174009799957275, 'time_between_consecutive_saves_sec': 22.569724321365356, 'checkpointer_blocking_start_time': 1777042567.0441759, 'checkpointer_blocking_duration_secs': 0.3412058353424072, 'get_old_steps_start_time': 1777042567.3853998, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1777042538.8667686, 'checkpoint_manager_blocking_duration_secs': 28.525742292404175} I0424 14:56:07.392658 133126962399040 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 14:56:07.392702 133126962399040 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:56:12.578044 133005154686720 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6 I0424 14:56:50.313215 133005163079424 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.766 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.96310639381409 s) (per-host) I0424 14:56:50.313333 133005163079424 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.927977s. I0424 14:56:56.255329 133005163079424 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.869956s. I0424 14:56:56.255620 132998687090432 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:56:56.255754 132998687090432 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:56:56.255812 132998687090432 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:56:56.257309 132998687090432 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 14:56:56.257483 133126962399040 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. I0424 14:56:56.257658 133126962399040 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:56:56.258667 133126962399040 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.888, Tokens/s/device: 13850.378, 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: Fri Apr 24 14:57:04 UTC 2026 EXIT_CODE=0