feat/nnx-set-defaults-trueXPK Start: Mon Apr 20 19:49:05 UTC 2026 2026-04-20 19:49:09.727953: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776714549.740603 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776714549.744301 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776714549.755753 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776714549.755771 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776714549.755773 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776714549.755775 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 19:49:28.898313: 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) I0420 19:49:29.418456 137744748050240 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 19:49:38,459:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 19:49:38.459614 137744748050240 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 19:49:38,461:jax._src.distributed:157: Connecting to JAX distributed service on mt-10-shardy-true-swgbp-slice-job-0-0.mt-10-shardy-true-swgbp:8482 I0420 19:49:38.461885 137744748050240 distributed.py:157] Connecting to JAX distributed service on mt-10-shardy-true-swgbp-slice-job-0-0.mt-10-shardy-true-swgbp:8482 I0420 19:49:39.931275 137744748050240 max_utils.py:284] Jax distributed system initialized! I0420 19:49:46.903486 137744748050240 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 19:49:46.903594 137744748050240 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 19:49:46.903636 137744748050240 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0420 19:49:46.903672 137744748050240 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 19:49:47.526985 137744748050240 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 19:49:47.527608 137744748050240 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 19:49:47.527794 137744748050240 checkpointing.py:688] Setting up checkpoint logger... I0420 19:49:47.527848 137744748050240 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 19:49:47.527893 137744748050240 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 19:49:47.528237 137744748050240 base_pytree_checkpoint_handler.py:415] 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 0x7d46b0f55190>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 19:49:50.471358 137744748050240 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 19:49:50.471751 137744748050240 checkpoint_manager.py:709] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d328428deb0>}, handler_registry=None I0420 19:49:50.471988 137744748050240 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d328428deb0>` 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`. I0420 19:49:50.472038 137744748050240 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3284292810>` 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`. I0420 19:49:50.472076 137744748050240 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 0x7d328428deb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d328428deb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3284292810>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3284292810>}). I0420 19:49:50.472598 137744748050240 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 19:49:50.472681 137744748050240 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d326c59c180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 19:49:51.182684 137744748050240 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints I0420 19:49:51.634654 137744748050240 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=10), 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/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d456e1912b0> I0420 19:49:51.634834 137744748050240 checkpointing.py:302] Checkpoint manager created! I0420 19:49:52.564198 137744748050240 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 19:49:52.564303 137744748050240 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:49:52.934585 137744748050240 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 19:49:52.934678 137744748050240 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:49:52.950367 137744748050240 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 19:49:52.950424 137744748050240 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:49:52.973160 137744748050240 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:49:52.973229 137744748050240 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:49:52.989050 137744748050240 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:49:52.989119 137744748050240 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:49:53.004970 137744748050240 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:49:53.005035 137744748050240 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:49:53.029646 137744748050240 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 19:49:53.029713 137744748050240 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:49:53.049071 137744748050240 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 19:49:53.049136 137744748050240 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 19:49:53.237588 137744748050240 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 19:49:53.237702 137744748050240 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 19:49:54.666408 137744748050240 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0420 19:49:54.666568 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 19:49:54.666634 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 19:49:54.666695 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0420 19:49:54.666750 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0420 19:49:54.666791 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0420 19:49:54.666845 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0420 19:49:54.666898 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0420 19:49:54.666939 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0420 19:49:54.666978 137744748050240 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0420 19:49:54.667027 137744748050240 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0420 19:49:54.667074 137744748050240 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0420 19:49:55.168066 137744748050240 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 19:49:55.168161 137744748050240 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 19:49:55.182986 137744748050240 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 19:49:55.183048 137744748050240 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 19:50:05.846921 137744748050240 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. I0420 19:50:05.847829 137744748050240 metric_logger.py:289] number parameters: 1.104 billion I0420 19:50:16.844381 137744748050240 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 19:50:17.011846 137744748050240 checkpointing.py:798] Waited 0.16746258735656738 seconds for step 0 to finish before starting checkpointing. I0420 19:50:17.014308 137744748050240 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:50:17.016180 137744748050240 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0420 19:50:17.018783 137744748050240 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0. I0420 19:50:17.603858 137744748050240 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 19:50:17.604918 137744748050240 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0420 19:50:17.604976 137744748050240 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 I0420 19:50:17.860410 137744748050240 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.256635s I0420 19:50:17.860594 137744748050240 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.012 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.512181282043457 s) (per-host) I0420 19:50:17.860651 137744748050240 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.512243s (batch_requests_ready=0.251981s, total_serialization_initiated=0.260192s, others=0.000070s) I0420 19:50:17.860740 137744748050240 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.516525s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.516506) I0420 19:50:17.861706 137620378900224 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0420 19:50:17.861836 137744748050240 async_checkpointer.py:561] Finished blocking save. Time taken: 0.845584s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0. I0420 19:50:18.194891 137744748050240 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:50:18.195266 137618796832512 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:50:18.195457 137744748050240 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776714617.014292, 'wait_for_prev_duration_secs': 6.341934204101562e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776714617.016217, 'checkpointer_blocking_duration_secs': 0.8458020687103271, 'get_old_steps_start_time': 1776714617.862038, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776714617.0124586, 'checkpoint_manager_blocking_duration_secs': 1.182943344116211} I0420 19:50:18.195567 137744748050240 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 19:50:18.195621 137744748050240 max_utils.py:750] Memstats: After params initialized: I0420 19:50:18.195685 137744748050240 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0420 19:50:18.195721 137744748050240 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0420 19:50:18.195750 137744748050240 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0420 19:50:18.195776 137744748050240 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0420 19:50:18.462071 137744748050240 metric_logger.py:185] completed step: 0, seconds: 10.996, TFLOP/s/device: 1.236, Tokens/s/device: 186.242, total_weights: 65536, loss: 10.874 I0420 19:50:18.635346 137744748050240 metric_logger.py:185] completed step: 1, seconds: 1.616, TFLOP/s/device: 8.406, Tokens/s/device: 1267.010, total_weights: 65536, loss: 10.874 I0420 19:50:19.037409 137744748050240 metric_logger.py:185] completed step: 2, seconds: 0.026, TFLOP/s/device: 532.265, Tokens/s/device: 80228.777, total_weights: 65536, loss: 10.263 I0420 19:50:19.185176 137744748050240 metric_logger.py:185] completed step: 3, seconds: 0.403, TFLOP/s/device: 33.744, Tokens/s/device: 5086.265, total_weights: 65536, loss: 9.734 I0420 19:50:19.480532 137744748050240 metric_logger.py:185] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.704, Tokens/s/device: 13370.503, total_weights: 65536, loss: 9.278 I0420 19:50:19.486404 137744748050240 metric_logger.py:185] completed step: 5, seconds: 0.148, TFLOP/s/device: 92.021, Tokens/s/device: 13870.452, total_weights: 65536, loss: 8.892 I0420 19:50:38.117088 137744748050240 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.926, Tokens/s/device: 6922.427, total_weights: 65536, loss: 8.592 I0420 19:50:38.264672 137744748050240 metric_logger.py:185] completed step: 7, seconds: 18.484, TFLOP/s/device: 0.735, Tokens/s/device: 110.800, total_weights: 65536, loss: 8.384 I0420 19:50:38.412204 137744748050240 metric_logger.py:185] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.144, Tokens/s/device: 13436.733, total_weights: 65536, loss: 8.255 I0420 19:50:38.559278 137744748050240 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 19:50:38.559950 137744748050240 checkpointing.py:798] Waited 0.0006923675537109375 seconds for step 9 to finish before starting checkpointing. I0420 19:50:38.562035 137744748050240 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:50:39.475719 2587 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 19:50:42.365323 137620387292928 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/items/array_metadatas/process_6 I0420 19:51:14.058113 137620378900224 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 27.854 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 56.70965600013733 s) (per-host) I0420 19:51:14.058235 137620378900224 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 56.196318s. I0420 19:51:21.950520 137620378900224 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 64.088586s. I0420 19:51:21.950803 137618796832512 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:51:21.950937 137618796832512 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:51:21.950987 137618796832512 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:51:21.952401 137618796832512 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 19:51:21.952602 137744748050240 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 19:51:21.952739 137744748050240 checkpoint_manager.py:1458] Waiting for previous save to complete took 43.390707 seconds. If this number is high, consider checkpointing less frequently. I0420 19:51:21.954633 137744748050240 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0420 19:51:21.958200 137744748050240 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9. I0420 19:51:22.488785 137744748050240 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0420 19:51:22.488883 137744748050240 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 I0420 19:51:22.536502 137744748050240 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.048898s I0420 19:51:22.536676 137744748050240 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.126 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3009300231933594 s) (per-host) I0420 19:51:22.536731 137744748050240 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.300993s (batch_requests_ready=0.250170s, total_serialization_initiated=0.050752s, others=0.000071s) I0420 19:51:22.536819 137744748050240 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.305011s (all_items=0.000015s, per_item={'items': '0.00001550'}, temp_paths=0.304996) I0420 19:51:22.537775 137619333703424 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0420 19:51:22.537944 137744748050240 async_checkpointer.py:561] Finished blocking save. Time taken: 0.583239s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9. I0420 19:51:22.542511 137744748050240 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:51:22.542793 137618796832512 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:51:22.542953 137744748050240 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776714638.5620065, 'wait_for_prev_duration_secs': 43.39070701599121, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776714681.9546723, 'checkpointer_blocking_duration_secs': 0.583378791809082, 'get_old_steps_start_time': 1776714682.5380683, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776714638.5601757, 'checkpoint_manager_blocking_duration_secs': 43.98274302482605} I0420 19:51:22.543057 137744748050240 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 19:51:22.543103 137744748050240 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:51:27.321519 137620387292928 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9/items/array_metadatas/process_6 I0420 19:52:04.495932 137619333703424 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 37.378 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.26014280319214 s) (per-host) I0420 19:52:04.496053 137619333703424 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.958165s. I0420 19:52:12.154797 137619333703424 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 49.616894s. I0420 19:52:12.155089 137618796832512 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:52:12.155205 137618796832512 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:52:12.155255 137618796832512 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:52:12.156855 137618796832512 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 19:52:12.157048 137744748050240 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 19:52:12.157213 137744748050240 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:52:12.157878 137744748050240 metric_logger.py:185] completed step: 9, seconds: 0.148, TFLOP/s/device: 92.099, Tokens/s/device: 13882.111, total_weights: 65536, loss: 8.179 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 19:52:22 UTC 2026 EXIT_CODE=0
XPK Start: Mon Apr 20 21:23:42 UTC 2026 2026-04-20 21:23:46.398360: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776720226.410929 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776720226.414638 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776720226.426136 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776720226.426157 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776720226.426159 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776720226.426161 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 21:24:05.527735: 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) I0420 21:24:06.055808 134799150839616 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 21:24:15,097:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 21:24:15.097308 134799150839616 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 21:24:15,099:jax._src.distributed:157: Connecting to JAX distributed service on mt-10-shardy-true-opucu-slice-job-0-0.mt-10-shardy-true-opucu:8482 I0420 21:24:15.099712 134799150839616 distributed.py:157] Connecting to JAX distributed service on mt-10-shardy-true-opucu-slice-job-0-0.mt-10-shardy-true-opucu:8482 I0420 21:24:16.192564 134799150839616 max_utils.py:284] Jax distributed system initialized! I0420 21:24:23.144876 134799150839616 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 21:24:23.144982 134799150839616 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 21:24:23.145022 134799150839616 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0420 21:24:23.145055 134799150839616 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 21:24:23.763612 134799150839616 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 21:24:23.876869 134799150839616 checkpointing.py:688] Setting up checkpoint logger... I0420 21:24:23.876987 134799150839616 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 21:24:23.877032 134799150839616 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 21:24:23.877252 134799150839616 base_pytree_checkpoint_handler.py:415] 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 0x7a98d1951a90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 21:24:26.815285 134799150839616 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 21:24:26.815527 134799150839616 checkpoint_manager.py:709] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a859c490c80>}, handler_registry=None I0420 21:24:26.815824 134799150839616 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a859c490c80>` 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`. I0420 21:24:26.815876 134799150839616 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a859c2d2810>` 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`. I0420 21:24:26.815911 134799150839616 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 0x7a859c490c80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a859c490c80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a859c2d2810>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a859c2d2810>}). I0420 21:24:26.816251 134799150839616 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 21:24:26.816330 134799150839616 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a847c494f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 21:24:27.637573 134799150839616 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints I0420 21:24:27.684423 134799150839616 checkpoint_manager.py:929] [process=0][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=10), 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_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a8454724680> I0420 21:24:27.684576 134799150839616 checkpointing.py:302] Checkpoint manager created! I0420 21:24:28.599387 134799150839616 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 21:24:28.599504 134799150839616 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 21:24:30.015755 134799150839616 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 21:24:30.015878 134799150839616 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:24:30.015934 134799150839616 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:24:30.015978 134799150839616 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 21:24:30.016017 134799150839616 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:24:30.016053 134799150839616 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:24:30.016103 134799150839616 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:24:30.016142 134799150839616 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 21:24:30.016178 134799150839616 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:24:30.016211 134799150839616 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:24:30.016244 134799150839616 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 21:24:30.016278 134799150839616 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 21:24:30.414591 134799150839616 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:24:30.414745 134799150839616 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:24:30.420131 134799150839616 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:24:30.420241 134799150839616 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:24:30.567934 134799150839616 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:24:30.568062 134799150839616 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:24:30.711305 134799150839616 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:24:30.711436 134799150839616 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:24:30.862635 134799150839616 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:24:30.862768 134799150839616 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:24:31.006169 134799150839616 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:24:31.006320 134799150839616 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:24:31.147685 134799150839616 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:24:31.147814 134799150839616 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:24:31.300467 134799150839616 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 21:24:31.300603 134799150839616 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:24:31.323530 134799150839616 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 21:24:31.323627 134799150839616 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 21:24:42.245611 134799150839616 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. I0420 21:24:42.253255 134799150839616 metric_logger.py:289] number parameters: 1.104 billion I0420 21:24:53.549994 134799150839616 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 21:24:53.700977 134799150839616 checkpointing.py:798] Waited 0.1509702205657959 seconds for step 0 to finish before starting checkpointing. I0420 21:24:53.703234 134799150839616 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:24:53.705054 134799150839616 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 0 I0420 21:24:53.708262 134799150839616 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0. I0420 21:24:54.403266 134799150839616 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 21:24:54.669349 134674001164032 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0 I0420 21:24:54.700953 134799150839616 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0420 21:24:54.701049 134799150839616 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 I0420 21:24:54.943597 134799150839616 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.244604s I0420 21:24:54.944540 134799150839616 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.895 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5327591896057129 s) (per-host) I0420 21:24:54.944614 134799150839616 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.532851s (batch_requests_ready=0.279842s, total_serialization_initiated=0.252143s, others=0.000867s) I0420 21:24:54.944698 134799150839616 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.537344s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.537325) I0420 21:24:54.945704 134673966470912 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 21:24:54.945861 134799150839616 async_checkpointer.py:561] Finished blocking save. Time taken: 1.240670s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0. I0420 21:24:54.985970 134799150839616 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:24:54.986285 134672390551296 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:24:54.986446 134799150839616 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776720293.7032168, 'wait_for_prev_duration_secs': 6.341934204101562e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776720293.7051592, 'checkpointer_blocking_duration_secs': 1.2408084869384766, 'get_old_steps_start_time': 1776720294.945989, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776720293.7014039, 'checkpoint_manager_blocking_duration_secs': 1.2850022315979004} I0420 21:24:54.986554 134799150839616 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 21:24:54.986630 134799150839616 max_utils.py:750] Memstats: After params initialized: I0420 21:24:54.986685 134799150839616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 21:24:54.986722 134799150839616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 21:24:54.986752 134799150839616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 21:24:54.986777 134799150839616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 21:24:55.294631 134799150839616 metric_logger.py:185] completed step: 0, seconds: 11.029, TFLOP/s/device: 1.232, Tokens/s/device: 185.687, total_weights: 65536, loss: 10.874 I0420 21:24:55.297381 134799150839616 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/tensorboard/' I0420 21:24:55.724963 134799150839616 metric_logger.py:185] completed step: 1, seconds: 1.741, TFLOP/s/device: 7.804, Tokens/s/device: 1176.314, total_weights: 65536, loss: 10.874 I0420 21:24:55.866976 134799150839616 metric_logger.py:185] completed step: 2, seconds: 0.432, TFLOP/s/device: 31.478, Tokens/s/device: 4744.772, total_weights: 65536, loss: 10.262 I0420 21:24:56.014591 134799150839616 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1166.077, Tokens/s/device: 175763.817, total_weights: 65536, loss: 9.732 I0420 21:25:14.887473 134673405576960 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776720296229042953, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776720296229042953, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0420 21:25:14.888594 134673984378624 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/items I0420 21:25:14.888999 134799150839616 metric_logger.py:185] completed step: 4, seconds: 0.144, TFLOP/s/device: 94.427, Tokens/s/device: 14233.095, total_weights: 65536, loss: 9.274 I0420 21:25:14.899691 134799150839616 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.309, Tokens/s/device: 13913.800, total_weights: 65536, loss: 8.888 I0420 21:25:15.043867 134799150839616 metric_logger.py:185] completed step: 6, seconds: 18.873, TFLOP/s/device: 0.720, Tokens/s/device: 108.515, total_weights: 65536, loss: 8.589 I0420 21:25:15.191202 134799150839616 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1604.526, Tokens/s/device: 241851.677, total_weights: 65536, loss: 8.380 I0420 21:25:15.338695 134799150839616 metric_logger.py:185] completed step: 8, seconds: 0.145, TFLOP/s/device: 93.655, Tokens/s/device: 14116.739, total_weights: 65536, loss: 8.251 I0420 21:25:15.487111 134799150839616 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 21:25:15.487744 134799150839616 checkpointing.py:798] Waited 0.0006616115570068359 seconds for step 9 to finish before starting checkpointing. I0420 21:25:15.489655 134799150839616 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:25:16.184261 2610 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 21:25:17.882182 134677172033280 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/items/array_metadatas/process_0 I0420 21:25:52.762157 134676685518592 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.575888s (commit=35.352028s, array_metadata_write=1.223860s) I0420 21:25:52.763323 134673966470912 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 27.070 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 58.35151124000549 s) (per-host) I0420 21:25:52.763447 134673966470912 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 57.817633s. I0420 21:25:53.519488 134673966470912 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776720296229042953, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0420 21:25:54.319171 134673405576960 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776720296229042953, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0420 21:25:54.494844 134673966470912 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000405s. I0420 21:25:55.029979 134673966470912 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 21:25:55.977614 134673966470912 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.336188s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/items I0420 21:25:55.980844 134673966470912 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0/items I0420 21:25:56.385986 134673966470912 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0 I0420 21:25:58.139811 134673966470912 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0`. I0420 21:25:58.140523 134673966470912 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 64.435343s. directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/0 I0420 21:25:58.141841 134673966470912 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 63.196027s. I0420 21:25:58.142050 134672390551296 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:25:58.142193 134672390551296 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:25:58.142260 134672390551296 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:25:58.145157 134672390551296 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 21:25:58.145341 134799150839616 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 21:25:58.145469 134799150839616 checkpoint_manager.py:1458] Waiting for previous save to complete took 42.655828 seconds. If this number is high, consider checkpointing less frequently. I0420 21:25:58.147577 134799150839616 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 9 I0420 21:25:58.150988 134799150839616 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9. I0420 21:25:58.682258 134673966470912 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9 I0420 21:25:58.696005 134799150839616 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0420 21:25:58.696115 134799150839616 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 I0420 21:25:58.724832 134799150839616 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.030882s I0420 21:25:58.725667 134799150839616 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.327 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2895646095275879 s) (per-host) I0420 21:25:58.725727 134799150839616 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.289641s (batch_requests_ready=0.252289s, total_serialization_initiated=0.036605s, others=0.000746s) I0420 21:25:58.725803 134799150839616 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.293547s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.293530) I0420 21:25:58.726638 134672901195520 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 21:25:58.726792 134799150839616 async_checkpointer.py:561] Finished blocking save. Time taken: 0.579148s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9. I0420 21:25:58.751702 134799150839616 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:25:58.751992 134672390551296 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:25:58.752142 134799150839616 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776720315.4896166, 'wait_for_prev_duration_secs': 42.65582776069641, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776720358.1476152, 'checkpointer_blocking_duration_secs': 0.579287052154541, 'get_old_steps_start_time': 1776720358.726925, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1776720315.4879718, 'checkpoint_manager_blocking_duration_secs': 43.26413631439209} I0420 21:25:58.752301 134799150839616 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 21:25:58.752346 134799150839616 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:26:01.419718 134676685518592 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9/items I0420 21:26:04.328138 134672917980928 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9/items/array_metadatas/process_0 I0420 21:26:45.126777 134672909588224 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.573195s (commit=41.028908s, array_metadata_write=1.544287s) I0420 21:26:45.127824 134672901195520 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 33.830 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.691688776016235 s) (per-host) I0420 21:26:45.127942 134672901195520 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.401190s. I0420 21:26:46.924526 134672901195520 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000369s. I0420 21:26:46.970615 134672901195520 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 21:26:47.615745 134672901195520 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.569691s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9/items I0420 21:26:47.617495 134672901195520 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9/items I0420 21:26:48.045977 134672901195520 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9 I0420 21:26:49.780951 134672901195520 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9`. I0420 21:26:49.781753 134672901195520 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 51.634112s. directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_10_shardy_true/checkpoints/9 I0420 21:26:49.783222 134672901195520 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.056471s. I0420 21:26:49.783432 134672390551296 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:26:49.783544 134672390551296 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:26:49.783611 134672390551296 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:26:49.786500 134672390551296 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 21:26:49.786689 134799150839616 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 21:26:49.786855 134799150839616 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:26:49.787495 134799150839616 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.455, Tokens/s/device: 13935.860, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 21:27:03 UTC 2026 EXIT_CODE=0