XPK Start: Mon Apr 20 18:43:59 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 18:44:22.995868: 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 18:44:23.176313 140719336675136 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 18:44:32.217882 140719336675136 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 18:44:32.220200 140719336675136 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-a8kvy-slice-job-0-0.mt-16-zarr3-true-sv-a8kvy:8482 I0420 18:44:33.311124 140719336675136 max_utils.py:284] Jax distributed system initialized! I0420 18:44:39.349263 140719336675136 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 18:44:39.349363 140719336675136 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 18:44:39.349403 140719336675136 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 I0420 18:44:39.349437 140719336675136 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 18:44:40.392740 140719336675136 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 18:44:40.507568 140719336675136 checkpointing.py:688] Setting up checkpoint logger... I0420 18:44:40.507680 140719336675136 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 18:44:40.507726 140719336675136 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 18:44:40.507926 140719336675136 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 0x7ffb2811c080>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 18:44:43.377993 140719336675136 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 18:44:43.378248 140719336675136 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ffa0273a360>}, handler_registry=None I0420 18:44:43.378487 140719336675136 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ffa0273a360>` 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 18:44:43.378536 140719336675136 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe754149a30>` 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 18:44:43.378572 140719336675136 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 0x7ffa0273a360>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ffa0273a360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe754149a30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe754149a30>}). I0420 18:44:43.378897 140719336675136 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 18:44:43.378970 140719336675136 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7fe7346cc4a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 18:44:44.583465 140719336675136 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints I0420 18:44:44.816233 140719336675136 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=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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ffa02b19e80> I0420 18:44:44.816399 140719336675136 checkpointing.py:302] Checkpoint manager created! I0420 18:44:45.764852 140719336675136 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 18:44:45.764966 140719336675136 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 18:44:47.185497 140719336675136 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 18:44:47.185612 140719336675136 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:44:47.185672 140719336675136 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:44:47.185718 140719336675136 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 18:44:47.185757 140719336675136 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:44:47.185794 140719336675136 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:44:47.185832 140719336675136 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:44:47.185868 140719336675136 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 18:44:47.185902 140719336675136 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:44:47.185935 140719336675136 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:44:47.185969 140719336675136 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 18:44:47.186002 140719336675136 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 18:44:47.588367 140719336675136 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:44:47.588461 140719336675136 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:44:47.593827 140719336675136 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:44:47.593881 140719336675136 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:44:47.610094 140719336675136 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:44:47.610154 140719336675136 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:44:47.625642 140719336675136 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:44:47.625699 140719336675136 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:44:47.649469 140719336675136 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:44:47.649543 140719336675136 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:44:47.665156 140719336675136 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:44:47.665213 140719336675136 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:44:47.680739 140719336675136 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:44:47.680802 140719336675136 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:44:47.711151 140719336675136 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 18:44:47.711222 140719336675136 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:44:47.731199 140719336675136 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 18:44:47.731262 140719336675136 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 18:44:58.892986 140719336675136 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 18:44:58.902746 140719336675136 metric_logger.py:301] number parameters: 1.104 billion I0420 18:45:10.290931 140719336675136 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 18:45:10.441991 140719336675136 checkpointing.py:798] Waited 0.15104365348815918 seconds for step 0 to finish before starting checkpointing. I0420 18:45:10.444244 140719336675136 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:45:10.445944 140719336675136 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 18:45:10.447316 140719336675136 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0. I0420 18:45:10.736609 140719336675136 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 18:45:10.757199 140719336675136 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:45:10.757270 140719336675136 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 18:45:11.024964 140719336675136 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269236s I0420 18:45:11.025874 140719336675136 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.504 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28024744987487793 s) (per-host) I0420 18:45:11.025940 140719336675136 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280334s (batch_requests_ready=0.003058s, total_serialization_initiated=0.276469s, others=0.000807s) I0420 18:45:11.026021 140719336675136 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285728s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.285711) I0420 18:45:11.026910 140719336675136 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.58 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0. I0420 18:45:11.027209 140592118331136 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:05:11.027184 I0420 18:45:11.040847 140719336675136 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:45:11.041108 140589408442112 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:45:11.041229 140719336675136 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710710.444225, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776710710.4459836, 'checkpointer_blocking_duration_secs': 0.5813238620758057, 'get_old_steps_start_time': 1776710711.0273318, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1776710710.4424129, 'checkpoint_manager_blocking_duration_secs': 0.5987772941589355} I0420 18:45:11.041335 140719336675136 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 18:45:11.041401 140719336675136 max_utils.py:750] Memstats: After params initialized: I0420 18:45:11.041450 140719336675136 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 18:45:11.041481 140719336675136 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 18:45:11.041507 140719336675136 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 18:45:11.041529 140719336675136 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 18:45:11.050072 140592088590080 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0 I0420 18:45:11.476577 140719336675136 metric_logger.py:196] completed step: 0, seconds: 11.081, TFLOP/s/device: 1.226, Tokens/s/device: 184.818, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0420 18:45:11.480110 140719336675136 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/tensorboard/' I0420 18:45:11.911618 140719336675136 metric_logger.py:196] completed step: 1, seconds: 1.182, TFLOP/s/device: 11.495, Tokens/s/device: 1732.649, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0420 18:45:12.053619 140719336675136 metric_logger.py:196] completed step: 2, seconds: 0.436, TFLOP/s/device: 31.164, Tokens/s/device: 4697.431, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0420 18:45:12.200814 140719336675136 metric_logger.py:196] completed step: 3, seconds: 0.013, TFLOP/s/device: 1081.175, Tokens/s/device: 162966.500, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0420 18:45:12.213292 140719336675136 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0420 18:45:12.495574 140719336675136 checkpointing.py:798] Waited 0.2822449207305908 seconds for step 5 to finish before starting checkpointing. I0420 18:45:12.498388 140719336675136 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:45:13.229749 140591552206592 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/items I0420 18:45:13.240198 140587298715392 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710712806651106, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776710712806651106, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:45:14.402830 2843 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 18:45:16.595646 140589928527616 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_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/items/array_metadatas/process_0 I0420 18:45:51.011460 140589416834816 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.620782s (commit=35.406476s, array_metadata_write=1.214306s) I0420 18:45:51.012809 140592118331136 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.228 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.267141580581665 s) (per-host) I0420 18:45:51.012978 140592118331136 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.985693s. I0420 18:45:51.785966 140592118331136 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710712806651106, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:45:52.509576 140592118331136 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000739s. I0420 18:45:52.560201 140587298715392 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710712806651106, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:45:52.914098 140592118331136 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:45:53.978363 140592118331136 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.068264s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/items I0420 18:45:53.980638 140592118331136 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0/items I0420 18:45:54.603991 140592118331136 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0 I0420 18:45:56.463174 140592118331136 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0`. I0420 18:45:56.464287 140592118331136 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.02 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/0 I0420 18:45:56.465920 140592118331136 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.438633s. I0420 18:45:56.466217 140589408442112 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:45:56.466373 140589408442112 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:45:56.466474 140589408442112 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:45:56.468972 140589408442112 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 18:45:56.469305 140719336675136 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 18:45:56.469485 140719336675136 checkpoint_manager.py:1452] Waiting for previous save to complete took 43.971106 seconds. If this number is high, consider checkpointing less frequently. I0420 18:45:56.471459 140719336675136 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0420 18:45:56.473864 140719336675136 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5. I0420 18:45:56.793991 140719336675136 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:45:56.794149 140719336675136 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 18:45:56.840803 140719336675136 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.049855s I0420 18:45:56.841775 140719336675136 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 22.307 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0691535472869873 s) (per-host) I0420 18:45:56.841889 140719336675136 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.069292s (batch_requests_ready=0.006150s, total_serialization_initiated=0.062269s, others=0.000873s) I0420 18:45:56.842031 140719336675136 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.074181s (all_items=0.000025s, per_item={'items': '0.00002503'}, temp_paths=0.074156) I0420 18:45:56.842975 140719336675136 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.37 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5. I0420 18:45:56.843448 140589408442112 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:05:56.843407 I0420 18:45:56.845858 140719336675136 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:45:56.846251 140590502381312 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:45:56.846480 140719336675136 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710712.4983358, 'wait_for_prev_duration_secs': 43.97110605239868, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776710756.4715378, 'checkpointer_blocking_duration_secs': 0.3721005916595459, 'get_old_steps_start_time': 1776710756.8436658, 'get_old_steps_duration_secs': 4.0531158447265625e-05, 'checkpoint_manager_blocking_start_time': 1776710712.4960475, 'checkpoint_manager_blocking_duration_secs': 44.350382566452026} I0420 18:45:56.846633 140719336675136 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0420 18:45:57.027255 140592118331136 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5 I0420 18:45:59.104612 140591056811776 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5/items I0420 18:46:02.085424 140591048419072 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_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5/items/array_metadatas/process_0 I0420 18:46:20.538388 140719336675136 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.772, Tokens/s/device: 14285.116, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0420 18:46:20.555295 140719336675136 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.530, Tokens/s/device: 13947.153, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0420 18:46:20.693233 140719336675136 metric_logger.py:196] completed step: 6, seconds: 68.342, TFLOP/s/device: 0.199, Tokens/s/device: 29.967, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0420 18:46:20.840858 140719336675136 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1581.554, Tokens/s/device: 238389.012, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0420 18:46:20.988654 140719336675136 metric_logger.py:196] completed step: 8, seconds: 0.140, TFLOP/s/device: 97.289, Tokens/s/device: 14664.499, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0420 18:46:21.136597 140719336675136 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 18:46:21.137275 140719336675136 checkpointing.py:798] Waited 0.0006899833679199219 seconds for step 9 to finish before starting checkpointing. I0420 18:46:21.139360 140719336675136 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:46:42.136534 140591040026368 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.890563s (commit=40.669186s, array_metadata_write=1.221377s) I0420 18:46:42.137754 140589408442112 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.820 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.365116357803345 s) (per-host) I0420 18:46:42.137881 140589408442112 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.294284s. I0420 18:46:43.575503 140589408442112 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000452s. I0420 18:46:43.963036 140589408442112 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:46:45.035918 140589408442112 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.067733s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5/items I0420 18:46:45.037879 140589408442112 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5/items I0420 18:46:45.602363 140589408442112 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5 I0420 18:46:47.034142 140589408442112 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5`. I0420 18:46:47.034841 140589408442112 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.56 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/5 I0420 18:46:47.036011 140589408442112 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.192411s. I0420 18:46:47.036211 140590502381312 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:46:47.036322 140590502381312 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:46:47.036390 140590502381312 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:46:47.038313 140590502381312 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0420 18:46:47.038492 140719336675136 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0420 18:46:47.038627 140719336675136 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.899268 seconds. If this number is high, consider checkpointing less frequently. I0420 18:46:47.040096 140719336675136 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0420 18:46:47.042075 140719336675136 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9. I0420 18:46:47.771529 140719336675136 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:46:47.771626 140719336675136 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 18:46:47.798929 140719336675136 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028872s I0420 18:46:47.799728 140719336675136 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.194 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03744649887084961 s) (per-host) I0420 18:46:47.799797 140719336675136 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037533s (batch_requests_ready=0.002607s, total_serialization_initiated=0.034211s, others=0.000715s) I0420 18:46:47.799882 140719336675136 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041640s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.041624) I0420 18:46:47.800642 140719336675136 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9. I0420 18:46:47.800920 140590502381312 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:06:47.800889 I0420 18:46:47.807796 140719336675136 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:46:47.808036 140591056811776 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:46:47.808156 140719336675136 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710781.1393254, 'wait_for_prev_duration_secs': 25.89926838874817, 'time_between_consecutive_saves_sec': 24.670283555984497, 'checkpointer_blocking_start_time': 1776710807.0401354, 'checkpointer_blocking_duration_secs': 0.7609133720397949, 'get_old_steps_start_time': 1776710807.8010678, 'get_old_steps_duration_secs': 4.029273986816406e-05, 'checkpoint_manager_blocking_start_time': 1776710781.1375127, 'checkpoint_manager_blocking_duration_secs': 26.670609712600708} I0420 18:46:47.808320 140719336675136 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 18:46:47.808365 140719336675136 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:46:48.018639 140589408442112 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9 I0420 18:46:49.850275 140591040026368 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9/items I0420 18:46:53.210295 140592118331136 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_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9/items/array_metadatas/process_0 I0420 18:47:33.563062 140591552206592 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.586115s (commit=41.338327s, array_metadata_write=1.247788s) I0420 18:47:33.564125 140590502381312 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.488 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.80180335044861 s) (per-host) I0420 18:47:33.564244 140590502381312 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.763217s. I0420 18:47:35.068915 140590502381312 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000398s. I0420 18:47:35.530121 140590502381312 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:47:36.565351 140590502381312 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.088193s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9/items I0420 18:47:36.566983 140590502381312 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9/items I0420 18:47:37.126522 140590502381312 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9 I0420 18:47:38.891229 140590502381312 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9`. I0420 18:47:38.892018 140590502381312 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.85 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_true/checkpoints/9 I0420 18:47:38.893545 140590502381312 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.092516s. I0420 18:47:38.893740 140591056811776 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:47:38.893850 140591056811776 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:47:38.893908 140591056811776 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:47:38.895375 140591056811776 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 18:47:38.895519 140719336675136 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 18:47:38.895653 140719336675136 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:47:38.896617 140719336675136 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.325, Tokens/s/device: 13916.163, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 18:47:50 UTC 2026 EXIT_CODE=0