XPK Start: Fri Apr 17 04:14:14 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 04:14:38.182340: 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) I0417 04:14:38.715349 138674169882432 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 04:14:47.756878 138674169882432 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 04:14:47.759312 138674169882432 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-o4tjs-slice-job-0-0.mt-15-ocdbt-false-rs-o4tjs:8482 I0417 04:14:49.363038 138674169882432 max_utils.py:284] Jax distributed system initialized! I0417 04:14:54.398792 138674169882432 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 04:14:54.398921 138674169882432 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 04:14:54.398962 138674169882432 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 I0417 04:14:54.398998 138674169882432 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 04:14:55.078773 138674169882432 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:14:55.079367 138674169882432 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:14:55.079546 138674169882432 checkpointing.py:688] Setting up checkpoint logger... I0417 04:14:55.079598 138674169882432 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0417 04:14:55.079640 138674169882432 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 04:14:55.079979 138674169882432 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7e1ef7deb620>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 04:14:56.866311 138674169882432 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 04:14:56.866544 138674169882432 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e0a98598920>}, handler_registry=None I0417 04:14:56.866782 138674169882432 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e0a98598920>` 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`. I0417 04:14:56.866847 138674169882432 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e0a9859ca40>` 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`. I0417 04:14:56.866883 138674169882432 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 0x7e0a98598920>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e0a98598920>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e0a9859ca40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e0a9859ca40>}). I0417 04:14:56.867199 138674169882432 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 04:14:56.867269 138674169882432 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e0a981c4680> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 04:14:58.728898 138555745613568 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776399163259059912, 'commit_timestamp_nsecs': 1776399179194927380, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/_CHECKPOINT_METADATA I0417 04:14:59.001041 138555754006272 checkpoint_manager.py:1871] Missing metrics for step 5 I0417 04:14:59.665071 138674169882432 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints I0417 04:15:01.603106 138674169882432 checkpoint_manager.py:929] [process=6][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e0a985998e0> I0417 04:15:01.603278 138674169882432 checkpointing.py:302] Checkpoint manager created! I0417 04:15:02.538039 138674169882432 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:15:02.538145 138674169882432 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:15:02.914959 138674169882432 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:15:02.915045 138674169882432 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:15:02.931893 138674169882432 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:15:02.931949 138674169882432 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:15:02.955750 138674169882432 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:15:02.955817 138674169882432 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:15:02.972420 138674169882432 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:15:02.972481 138674169882432 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:15:02.989053 138674169882432 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:15:02.989110 138674169882432 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:15:03.014961 138674169882432 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 04:15:03.015029 138674169882432 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:15:03.036197 138674169882432 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 04:15:03.036258 138674169882432 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 04:15:03.255584 138674169882432 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 04:15:03.255688 138674169882432 checkpointing.py:582] restoring from this run's directory step 9 I0417 04:15:03.512462 138674169882432 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9. I0417 04:15:03.760336 138674169882432 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9. I0417 04:15:03.760794 138674169882432 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9. I0417 04:15:07.632852 1440 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 04:15:18.048320 138674169882432 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 135.558 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.65757417678833 s) (per-host) I0417 04:15:18.056034 138674169882432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 126.698 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 12.491063356399536 s) (per-host) I0417 04:15:18.132370 138674169882432 event_tracking.py:138] [process=6] [sync] Finished load in 14.62 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9 I0417 04:15:18.132519 138674169882432 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776399303.5123682, 'checkpointer_duration_secs': 14.62012004852295, 'checkpoint_manager_start_time': 1776399303.2559721, 'checkpoint_manager_duration_secs': 14.87651801109314} fsdp: 32 I0417 04:15:18.550521 138674169882432 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 04:15:18.550657 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 04:15:18.550711 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 04:15:18.550768 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 04:15:18.550831 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 04:15:18.550888 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 04:15:18.550948 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 04:15:18.551002 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0417 04:15:18.551041 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0417 04:15:18.551077 138674169882432 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 04:15:18.551123 138674169882432 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 04:15:18.551168 138674169882432 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 04:15:19.040701 138674169882432 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0417 04:15:19.040812 138674169882432 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0417 04:15:19.057025 138674169882432 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0417 04:15:19.057083 138674169882432 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0417 04:15:30.098633 138674169882432 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. I0417 04:15:30.099435 138674169882432 metric_logger.py:289] number parameters: 1.104 billion I0417 04:15:41.650020 138674169882432 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 04:15:41.804785 138674169882432 checkpointing.py:798] Waited 0.15474224090576172 seconds for step 10 to finish before starting checkpointing. I0417 04:15:41.807339 138674169882432 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:15:41.809049 138674169882432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0417 04:15:41.810868 138674169882432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/10. I0417 04:15:42.531205 138674169882432 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 04:15:42.532167 138674169882432 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:15:42.532222 138674169882432 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 I0417 04:15:42.808016 138674169882432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.276883s I0417 04:15:42.808178 138674169882432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.493 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2808072566986084 s) (per-host) I0417 04:15:42.808234 138674169882432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280872s (batch_requests_ready=0.002228s, total_serialization_initiated=0.278570s, others=0.000074s) I0417 04:15:42.808324 138674169882432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285232s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.285215) I0417 04:15:42.809056 138674169882432 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.00 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/10. I0417 04:15:42.809326 138552977364736 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:35:42.809300 I0417 04:15:42.815023 138674169882432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:15:42.815252 138551384270592 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:15:42.815364 138674169882432 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776399341.8073213, 'wait_for_prev_duration_secs': 5.91278076171875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776399341.809087, 'checkpointer_blocking_duration_secs': 1.0003464221954346, 'get_old_steps_start_time': 1776399342.809456, 'get_old_steps_duration_secs': 3.3855438232421875e-05, 'checkpoint_manager_blocking_start_time': 1776399341.80533, 'checkpoint_manager_blocking_duration_secs': 1.010000467300415} I0417 04:15:42.815458 138674169882432 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 04:15:42.815503 138674169882432 max_utils.py:750] Memstats: After params initialized: I0417 04:15:42.815547 138674169882432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0417 04:15:42.815577 138674169882432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0417 04:15:42.815602 138674169882432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0417 04:15:42.815624 138674169882432 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0417 04:15:43.127798 138674169882432 metric_logger.py:185] completed step: 10, seconds: 11.550, TFLOP/s/device: 1.176, Tokens/s/device: 177.309, total_weights: 65536, loss: 6.833 I0417 04:15:43.302373 138674169882432 metric_logger.py:185] completed step: 11, seconds: 1.477, TFLOP/s/device: 9.202, Tokens/s/device: 1387.051, total_weights: 65536, loss: 6.527 I0417 04:15:43.449891 138674169882432 metric_logger.py:185] completed step: 12, seconds: 0.028, TFLOP/s/device: 494.059, Tokens/s/device: 74470.019, total_weights: 65536, loss: 6.265 I0417 04:15:43.597219 138674169882432 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.984, Tokens/s/device: 13412.621, total_weights: 65536, loss: 6.045 I0417 04:15:43.601606 138674169882432 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 04:15:43.891606 138674169882432 checkpointing.py:798] Waited 0.289963960647583 seconds for step 15 to finish before starting checkpointing. I0417 04:15:43.894536 138674169882432 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:15:48.956192 138551392663296 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_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_6 I0417 04:15:53.064665 138552977364736 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 149.906 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.537251472473145 s) (per-host) I0417 04:15:53.064792 138552977364736 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.255389s. I0417 04:16:00.975390 138552977364736 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 18.165972s. I0417 04:16:00.975620 138551384270592 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:16:00.975676 138551384270592 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:16:00.975727 138551384270592 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:16:00.977197 138551384270592 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 04:16:00.977355 138674169882432 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0417 04:16:00.977488 138674169882432 checkpoint_manager.py:1452] Waiting for previous save to complete took 17.082948 seconds. If this number is high, consider checkpointing less frequently. I0417 04:16:00.979383 138674169882432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0417 04:16:00.981851 138674169882432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/15. I0417 04:16:01.687037 138674169882432 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:16:01.687129 138674169882432 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 I0417 04:16:01.712957 138674169882432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027000s I0417 04:16:01.713091 138674169882432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 49.478 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031177043914794922 s) (per-host) I0417 04:16:01.713139 138674169882432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031233s (batch_requests_ready=0.002569s, total_serialization_initiated=0.028601s, others=0.000062s) I0417 04:16:01.713213 138674169882432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.036142s (all_items=0.000013s, per_item={'items': '0.00001287'}, temp_paths=0.036129) I0417 04:16:01.713878 138674169882432 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/15. I0417 04:16:01.714224 138551384270592 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:36:01.714184 I0417 04:16:01.724152 138674169882432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:16:01.724476 138547608655616 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:16:01.724643 138674169882432 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776399343.8945074, 'wait_for_prev_duration_secs': 17.082948446273804, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776399360.9794211, 'checkpointer_blocking_duration_secs': 0.734952449798584, 'get_old_steps_start_time': 1776399361.714395, 'get_old_steps_duration_secs': 3.409385681152344e-05, 'checkpoint_manager_blocking_start_time': 1776399343.89226, 'checkpoint_manager_blocking_duration_secs': 17.832348108291626} I0417 04:16:01.724756 138674169882432 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 04:16:01.725387 138674169882432 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.088, Tokens/s/device: 13880.511, total_weights: 65536, loss: 5.864 I0417 04:16:01.731170 138674169882432 metric_logger.py:185] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.280, Tokens/s/device: 13909.453, total_weights: 65536, loss: 5.719 I0417 04:16:06.932209 138551392663296 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_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_6 I0417 04:16:12.358030 138551384270592 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 147.957 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.676080226898193 s) (per-host) I0417 04:16:12.358138 138551384270592 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.643795s. I0417 04:16:17.873831 138551384270592 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 16.159454s. I0417 04:16:17.874148 138547608655616 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:16:17.874263 138547608655616 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:16:17.874311 138547608655616 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:16:17.875760 138547608655616 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 04:16:23.405883 138674169882432 metric_logger.py:185] completed step: 16, seconds: 18.128, TFLOP/s/device: 0.750, Tokens/s/device: 112.973, total_weights: 65536, loss: 5.605 I0417 04:16:23.553816 138674169882432 metric_logger.py:185] completed step: 17, seconds: 21.529, TFLOP/s/device: 0.631, Tokens/s/device: 95.128, total_weights: 65536, loss: 5.517 I0417 04:16:23.700927 138674169882432 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.421, Tokens/s/device: 13478.473, total_weights: 65536, loss: 5.448 I0417 04:16:23.848123 138674169882432 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 04:16:23.848762 138674169882432 checkpointing.py:798] Waited 0.0006630420684814453 seconds for step 19 to finish before starting checkpointing. I0417 04:16:23.851041 138674169882432 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:16:23.852710 138674169882432 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0417 04:16:23.854169 138674169882432 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/19. I0417 04:16:24.130000 138674169882432 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:16:24.130088 138674169882432 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 I0417 04:16:24.155344 138674169882432 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026197s I0417 04:16:24.155467 138674169882432 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 51.985 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.029673337936401367 s) (per-host) I0417 04:16:24.155512 138674169882432 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.029727s (batch_requests_ready=0.001748s, total_serialization_initiated=0.027921s, others=0.000058s) I0417 04:16:24.155586 138674169882432 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034003s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.033992) I0417 04:16:24.156266 138674169882432 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.30 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/19. I0417 04:16:24.156588 138547608655616 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:36:24.156547 I0417 04:16:24.580191 138674169882432 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:16:24.580580 138546576877312 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:16:24.580727 138674169882432 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776399383.8510106, 'wait_for_prev_duration_secs': 6.890296936035156e-05, 'time_between_consecutive_saves_sec': 5.9752113819122314, 'checkpointer_blocking_start_time': 1776399383.8527474, 'checkpointer_blocking_duration_secs': 0.3039863109588623, 'get_old_steps_start_time': 1776399384.1567523, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776399383.8490233, 'checkpoint_manager_blocking_duration_secs': 0.7316699028015137} I0417 04:16:24.580940 138674169882432 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 04:16:24.580989 138674169882432 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:16:29.327037 138551392663296 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_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_6 I0417 04:16:34.484661 138547608655616 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 152.488 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.358835220336914 s) (per-host) I0417 04:16:34.484771 138547608655616 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.328072s. I0417 04:16:40.196337 138547608655616 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 16.039623s. I0417 04:16:40.196604 138546576877312 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:16:40.196725 138546576877312 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:16:40.196765 138546576877312 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:16:40.198278 138546576877312 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 04:16:40.198446 138674169882432 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0417 04:16:40.198592 138674169882432 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:16:40.199216 138674169882432 metric_logger.py:185] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.829, Tokens/s/device: 13841.485, total_weights: 65536, loss: 5.391 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 04:16:51 UTC 2026 EXIT_CODE=0