XPK Start: Fri Apr 24 14:43:04 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 14:43:29.729086: 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 14:43:29.944939 134238768506688 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:43:38.987187 134238768506688 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:43:38.989564 134238768506688 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-rs-p5gyv-slice-job-0-0.mt-16-zarr3-true-rs-p5gyv:8482 I0424 14:43:39.967694 134238768506688 max_utils.py:284] Jax distributed system initialized! I0424 14:43:46.108384 134238768506688 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:43:46.108490 134238768506688 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:43:46.108533 134238768506688 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 14:43:46.108568 134238768506688 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:43:46.806086 134238768506688 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:43:46.918758 134238768506688 checkpointing.py:688] Setting up checkpoint logger... I0424 14:43:46.918878 134238768506688 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 14:43:46.918923 134238768506688 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:43:46.919127 134238768506688 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 0x7a162e548dd0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:43:48.276709 134238768506688 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:43:48.276944 134238768506688 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 0x7a02b82389b0>}, handler_registry=None I0424 14:43:48.277218 134238768506688 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a02b82389b0>` 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 14:43:48.277274 134238768506688 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02b8235bb0>` 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 14:43:48.277310 134238768506688 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 0x7a02b82389b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a02b82389b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02b8235bb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02b8235bb0>}). I0424 14:43:48.277641 134238768506688 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:43:48.277720 134238768506688 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a02884bf560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:43:50.058014 134120621471488 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777041616462965236, 'commit_timestamp_nsecs': 1777041663330144030, '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_true/checkpoints/5/_CHECKPOINT_METADATA I0424 14:43:50.303327 134120613078784 checkpoint_manager.py:1871] Missing metrics for step 9 I0424 14:43:50.313479 134238768506688 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_true/checkpoints I0424 14:43:50.583417 134238768506688 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_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a02b8238da0> I0424 14:43:50.583595 134238768506688 checkpointing.py:302] Checkpoint manager created! I0424 14:43:51.557118 134238768506688 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:43:51.557236 134238768506688 checkpointing.py:582] restoring from this run's directory step 9 I0424 14:43:51.817725 134238768506688 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_true/checkpoints/9. I0424 14:43:52.073388 134238768506688 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_true/checkpoints/9. I0424 14:43:52.073824 134238768506688 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_true/checkpoints/9. I0424 14:43:53.827524 1416 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:44:01.257895 134238768506688 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 212.490 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.436960458755493 s) (per-host) I0424 14:44:01.267066 134238768506688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 201.252 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.863788843154907 s) (per-host) I0424 14:44:01.708660 134238768506688 event_tracking.py:138] [process=6] [sync] Finished load in 9.89 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_true/checkpoints/9 I0424 14:44:01.708819 134238768506688 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_true/checkpoints', 'checkpointer_start_time': 1777041831.8176277, 'checkpointer_duration_secs': 9.891158819198608, 'checkpoint_manager_start_time': 1777041831.5578437, 'checkpoint_manager_duration_secs': 10.150944948196411} fsdp: 32 I0424 14:44:01.785150 134238768506688 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:44:01.785269 134238768506688 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:44:01.785349 134238768506688 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:44:01.785427 134238768506688 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:44:01.785489 134238768506688 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:44:01.785539 134238768506688 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:44:01.785609 134238768506688 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:44:01.785674 134238768506688 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:44:01.785738 134238768506688 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:44:01.785793 134238768506688 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:44:01.785847 134238768506688 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:44:01.785897 134238768506688 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:44:02.200764 134238768506688 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:44:02.200859 134238768506688 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:44:02.206304 134238768506688 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:44:02.206360 134238768506688 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:44:02.222877 134238768506688 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:44:02.222942 134238768506688 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:44:02.238812 134238768506688 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:44:02.238880 134238768506688 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:44:02.264120 134238768506688 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 14:44:02.264185 134238768506688 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:44:02.280266 134238768506688 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 14:44:02.280326 134238768506688 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:44:02.296362 134238768506688 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 14:44:02.296422 134238768506688 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:44:02.326041 134238768506688 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:44:02.326129 134238768506688 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:44:02.346147 134238768506688 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:44:02.346207 134238768506688 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:44:13.859475 134238768506688 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 14:44:13.861811 134238768506688 metric_logger.py:301] number parameters: 1.104 billion I0424 14:44:25.364154 134238768506688 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 14:44:25.515856 134238768506688 checkpointing.py:798] Waited 0.1516876220703125 seconds for step 10 to finish before starting checkpointing. I0424 14:44:25.518435 134238768506688 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:44:25.520479 134238768506688 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0424 14:44:25.521953 134238768506688 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_true/checkpoints/10. I0424 14:44:26.194621 134238768506688 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:44:26.196798 134238768506688 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:44:26.196870 134238768506688 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 14:44:26.479412 134238768506688 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.285585s I0424 14:44:26.479582 134238768506688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.272 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29257893562316895 s) (per-host) I0424 14:44:26.479639 134238768506688 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.292646s (batch_requests_ready=0.002692s, total_serialization_initiated=0.289879s, others=0.000075s) I0424 14:44:26.479735 134238768506688 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.297181s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.297162) I0424 14:44:26.480473 134238768506688 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.96 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_true/checkpoints/10. I0424 14:44:26.480825 134119530944256 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:04:26.480787 I0424 14:44:26.484776 134238768506688 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:44:26.485015 134116857263872 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:44:26.485161 134238768506688 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041865.5184166, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041865.5205195, 'checkpointer_blocking_duration_secs': 0.9604558944702148, 'get_old_steps_start_time': 1777041866.4809988, 'get_old_steps_duration_secs': 3.504753112792969e-05, 'checkpoint_manager_blocking_start_time': 1777041865.5162482, 'checkpoint_manager_blocking_duration_secs': 0.9688756465911865} I0424 14:44:26.485315 134238768506688 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 14:44:26.485385 134238768506688 max_utils.py:750] Memstats: After params initialized: I0424 14:44:26.485433 134238768506688 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 14:44:26.485465 134238768506688 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 14:44:26.485491 134238768506688 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 14:44:26.485515 134238768506688 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 14:44:26.810585 134238768506688 metric_logger.py:196] completed step: 10, seconds: 11.501, TFLOP/s/device: 1.181, Tokens/s/device: 178.068, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0424 14:44:26.988044 134238768506688 metric_logger.py:196] completed step: 11, seconds: 1.445, TFLOP/s/device: 9.405, Tokens/s/device: 1417.654, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0424 14:44:27.135798 134238768506688 metric_logger.py:196] completed step: 12, seconds: 0.020, TFLOP/s/device: 675.943, Tokens/s/device: 101885.478, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0424 14:44:27.283399 134238768506688 metric_logger.py:196] completed step: 13, seconds: 0.166, TFLOP/s/device: 81.733, Tokens/s/device: 12319.686, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0424 14:44:27.289709 134238768506688 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 14:44:27.578506 134238768506688 checkpointing.py:798] Waited 0.2887687683105469 seconds for step 15 to finish before starting checkpointing. I0424 14:44:27.581342 134238768506688 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:44:31.406739 134117349889792 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_true/checkpoints/10/items/array_metadatas/process_6 I0424 14:45:08.189441 134119530944256 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.607 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.00240111351013 s) (per-host) I0424 14:45:08.189562 134119530944256 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.708625s. I0424 14:45:16.499503 134119530944256 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.018549s. I0424 14:45:16.499777 134116857263872 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:45:16.499907 134116857263872 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:45:16.499955 134116857263872 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:45:16.501522 134116857263872 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 14:45:16.501670 134238768506688 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 14:45:16.501803 134238768506688 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.920461 seconds. If this number is high, consider checkpointing less frequently. I0424 14:45:16.504077 134238768506688 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0424 14:45:16.506755 134238768506688 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_true/checkpoints/15. I0424 14:45:16.834737 134238768506688 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:45:16.834828 134238768506688 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 14:45:16.861453 134238768506688 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028373s I0424 14:45:16.861625 134238768506688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 43.589 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03538870811462402 s) (per-host) I0424 14:45:16.861678 134238768506688 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035450s (batch_requests_ready=0.002719s, total_serialization_initiated=0.032663s, others=0.000068s) I0424 14:45:16.861782 134238768506688 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040568s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.040556) I0424 14:45:16.862521 134238768506688 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.36 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_true/checkpoints/15. I0424 14:45:16.862834 134116857263872 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:05:16.862800 I0424 14:45:16.885083 134238768506688 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:45:16.885364 134117349889792 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:45:16.885508 134238768506688 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041867.581314, 'wait_for_prev_duration_secs': 48.92046070098877, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041916.5041325, 'checkpointer_blocking_duration_secs': 0.35883450508117676, 'get_old_steps_start_time': 1777041916.8629916, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1777041867.5788672, 'checkpoint_manager_blocking_duration_secs': 49.30660152435303} I0424 14:45:16.885621 134238768506688 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 14:45:16.886778 134238768506688 metric_logger.py:196] completed step: 14, seconds: 0.146, TFLOP/s/device: 92.850, Tokens/s/device: 13995.380, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0424 14:45:16.894804 134238768506688 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.837, Tokens/s/device: 13842.608, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0424 14:45:22.117803 134114699106048 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_true/checkpoints/15/items/array_metadatas/process_6 I0424 14:45:38.738757 134238768506688 metric_logger.py:196] completed step: 16, seconds: 49.603, TFLOP/s/device: 0.274, Tokens/s/device: 41.288, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0424 14:45:38.886143 134238768506688 metric_logger.py:196] completed step: 17, seconds: 21.699, TFLOP/s/device: 0.626, Tokens/s/device: 94.382, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0424 14:45:39.033662 134238768506688 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.846, Tokens/s/device: 13391.835, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0424 14:45:39.181828 134238768506688 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 14:45:39.182523 134238768506688 checkpointing.py:798] Waited 0.0007052421569824219 seconds for step 19 to finish before starting checkpointing. I0424 14:45:39.184450 134238768506688 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:45:57.928487 134116857263872 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.431 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.102216720581055 s) (per-host) I0424 14:45:57.928596 134116857263872 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.065655s. I0424 14:46:06.781588 134116857263872 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.918630s. I0424 14:46:06.781874 134117349889792 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:46:06.781993 134117349889792 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:46:06.782055 134117349889792 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:46:06.783464 134117349889792 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 14:46:06.783645 134238768506688 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 14:46:06.783768 134238768506688 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.599327 seconds. If this number is high, consider checkpointing less frequently. I0424 14:46:06.785469 134238768506688 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0424 14:46:06.787563 134238768506688 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_true/checkpoints/19. I0424 14:46:07.095409 134238768506688 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:46:07.095503 134238768506688 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 14:46:07.121485 134238768506688 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027603s I0424 14:46:07.121656 134238768506688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 44.399 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03474307060241699 s) (per-host) I0424 14:46:07.121707 134238768506688 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.034804s (batch_requests_ready=0.002882s, total_serialization_initiated=0.031857s, others=0.000066s) I0424 14:46:07.121806 134238768506688 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038864s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.038852) I0424 14:46:07.122596 134238768506688 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_true/checkpoints/19. I0424 14:46:07.122911 134117349889792 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:06:07.122876 I0424 14:46:07.126994 134238768506688 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:46:07.127248 134114699106048 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:46:07.127392 134238768506688 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041939.1844199, 'wait_for_prev_duration_secs': 27.599326610565186, 'time_between_consecutive_saves_sec': 22.682859897613525, 'checkpointer_blocking_start_time': 1777041966.7855103, 'checkpointer_blocking_duration_secs': 0.3375415802001953, 'get_old_steps_start_time': 1777041967.123071, 'get_old_steps_duration_secs': 3.8623809814453125e-05, 'checkpoint_manager_blocking_start_time': 1777041939.1827552, 'checkpoint_manager_blocking_duration_secs': 27.94460368156433} I0424 14:46:07.127508 134238768506688 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 14:46:07.127550 134238768506688 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:46:12.296896 134116848871168 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_true/checkpoints/19/items/array_metadatas/process_6 I0424 14:46:48.355911 134117349889792 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.276 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.26896095275879 s) (per-host) I0424 14:46:48.356033 134117349889792 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.233012s. I0424 14:46:55.806752 134117349889792 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.683716s. I0424 14:46:55.807036 134114699106048 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:46:55.807168 134114699106048 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:46:55.807209 134114699106048 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:46:55.808595 134114699106048 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 14:46:55.811007 134238768506688 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 14:46:55.811198 134238768506688 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:46:55.812055 134238768506688 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.276, Tokens/s/device: 13908.791, 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 14:47:04 UTC 2026 EXIT_CODE=0