XPK Start: Tue Apr 21 12:54:35 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-21 12:54:59.902477: 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) I0421 12:55:00.082308 137843719460672 max_utils.py:273] Attempting to initialize the jax distributed system... I0421 12:55:09.125765 137843719460672 distributed.py:149] Starting JAX distributed service on [::]:8482 I0421 12:55:09.128117 137843719460672 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-j0q17-slice-job-0-0.mt-15-ocdbt-false-rs-j0q17:8482 I0421 12:55:10.152717 137843719460672 max_utils.py:284] Jax distributed system initialized! I0421 12:55:15.505871 137843719460672 max_utils.py:800] System Information: Jax Version: 0.9.2 I0421 12:55:15.505972 137843719460672 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0421 12:55:15.506012 137843719460672 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 I0421 12:55:15.506046 137843719460672 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 12:55:16.206455 137843719460672 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 12:55:16.207048 137843719460672 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 12:55:16.207228 137843719460672 checkpointing.py:688] Setting up checkpoint logger... I0421 12:55:16.207280 137843719460672 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0421 12:55:16.207322 137843719460672 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0421 12:55:16.207652 137843719460672 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 0x7d5d9cf1d1f0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 12:55:17.487131 137843719460672 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0421 12:55:17.487367 137843719460672 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d49e8542e40>}, handler_registry=None I0421 12:55:17.487611 137843719460672 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d49e8542e40>` 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`. I0421 12:55:17.487662 137843719460672 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d49e83e99a0>` 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`. I0421 12:55:17.487699 137843719460672 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 0x7d49e8542e40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d49e8542e40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d49e83e99a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d49e83e99a0>}). I0421 12:55:17.488045 137843719460672 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0421 12:55:17.488119 137843719460672 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d48e477c540> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0421 12:55:19.482937 137724828837632 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776775966801903319, 'commit_timestamp_nsecs': 1776775982314594579, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/5/_CHECKPOINT_METADATA I0421 12:55:19.719660 137724820444928 checkpoint_manager.py:1871] Missing metrics for step 9 I0421 12:55:20.154474 137843719460672 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints I0421 12:55:21.048849 137843719460672 checkpoint_manager.py:929] [process=7][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_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d49e83e7650> I0421 12:55:21.049023 137843719460672 checkpointing.py:302] Checkpoint manager created! I0421 12:55:21.984012 137843719460672 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 12:55:21.984117 137843719460672 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:55:22.364550 137843719460672 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 12:55:22.364652 137843719460672 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:55:22.381060 137843719460672 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 12:55:22.381116 137843719460672 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:55:22.404978 137843719460672 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:55:22.405050 137843719460672 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:55:22.421594 137843719460672 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:55:22.421659 137843719460672 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:55:22.438058 137843719460672 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:55:22.438116 137843719460672 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:55:22.463016 137843719460672 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 12:55:22.463084 137843719460672 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:55:22.485051 137843719460672 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 12:55:22.485117 137843719460672 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 12:55:22.701496 137843719460672 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0421 12:55:22.701611 137843719460672 checkpointing.py:582] restoring from this run's directory step 9 I0421 12:55:22.953729 137843719460672 event_tracking.py:70] [process=7] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/9. I0421 12:55:23.197200 137843719460672 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/9. I0421 12:55:23.197630 137843719460672 event_tracking.py:125] [process=7] [sync] Finished blocking load in 0.24 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/9. I0421 12:55:25.812672 1443 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 12:55:34.455849 137843719460672 jax_array_handlers.py:843] [process=7] /jax/orbax/read/worker/io/requested throughput: 163.020 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 9.693723440170288 s) (per-host) I0421 12:55:34.462246 137843719460672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/checkpoint/read/gbytes_per_sec: 156.575 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.107585906982422 s) (per-host) I0421 12:55:36.528301 137843719460672 event_tracking.py:138] [process=7] [sync] Finished load in 13.57 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/9 I0421 12:55:36.528454 137843719460672 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776776122.9536314, 'checkpointer_duration_secs': 13.574790239334106, 'checkpoint_manager_start_time': 1776776122.7019036, 'checkpoint_manager_duration_secs': 13.826520919799805} fsdp: 32 I0421 12:55:36.942940 137843719460672 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0421 12:55:36.943066 137843719460672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 12:55:36.943142 137843719460672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 12:55:36.943216 137843719460672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 12:55:36.943276 137843719460672 maxtext_utils.py:1827] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0421 12:55:36.943315 137843719460672 maxtext_utils.py:1827] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0421 12:55:36.943368 137843719460672 maxtext_utils.py:1827] 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) I0421 12:55:36.943418 137843719460672 maxtext_utils.py:1827] 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') I0421 12:55:36.943458 137843719460672 maxtext_utils.py:1827] 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) I0421 12:55:36.943493 137843719460672 maxtext_utils.py:1827] 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) I0421 12:55:36.943539 137843719460672 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 12:55:36.943584 137843719460672 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 12:55:37.432326 137843719460672 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 12:55:37.432417 137843719460672 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0421 12:55:37.448042 137843719460672 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 12:55:37.448101 137843719460672 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0421 12:55:48.398299 137843719460672 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. I0421 12:55:48.399069 137843719460672 metric_logger.py:301] number parameters: 1.104 billion I0421 12:55:59.940968 137843719460672 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0421 12:56:00.095120 137843719460672 checkpointing.py:798] Waited 0.15412640571594238 seconds for step 10 to finish before starting checkpointing. I0421 12:56:00.097473 137843719460672 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 12:56:00.099374 137843719460672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 10 I0421 12:56:00.100787 137843719460672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/10. I0421 12:56:00.451221 137843719460672 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 12:56:00.452341 137843719460672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 12:56:00.452397 137843719460672 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 I0421 12:56:00.733074 137843719460672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.281931s I0421 12:56:00.733238 137843719460672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.399 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28572678565979004 s) (per-host) I0421 12:56:00.733289 137843719460672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285788s (batch_requests_ready=0.002093s, total_serialization_initiated=0.283626s, others=0.000070s) I0421 12:56:00.733372 137843719460672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289916s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.289898) I0421 12:56:00.734040 137843719460672 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.63 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/10. I0421 12:56:00.734378 137722110912256 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:16:00.734339 I0421 12:56:00.744440 137843719460672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 12:56:00.744769 137719996999424 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 12:56:00.744924 137843719460672 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776160.0974545, 'wait_for_prev_duration_secs': 5.9604644775390625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776776160.0994127, 'checkpointer_blocking_duration_secs': 0.6351182460784912, 'get_old_steps_start_time': 1776776160.7345533, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1776776160.0956154, 'checkpoint_manager_blocking_duration_secs': 0.6492722034454346} I0421 12:56:00.745027 137843719460672 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0421 12:56:00.745079 137843719460672 max_utils.py:750] Memstats: After params initialized: I0421 12:56:00.745131 137843719460672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0421 12:56:00.745165 137843719460672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0421 12:56:00.745193 137843719460672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0421 12:56:00.745218 137843719460672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0421 12:56:01.060017 137843719460672 metric_logger.py:196] completed step: 10, seconds: 11.542, TFLOP/s/device: 1.177, Tokens/s/device: 177.442, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0421 12:56:01.242762 137843719460672 metric_logger.py:196] completed step: 11, seconds: 1.117, TFLOP/s/device: 12.159, Tokens/s/device: 1832.723, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0421 12:56:01.390268 137843719460672 metric_logger.py:196] completed step: 12, seconds: 0.036, TFLOP/s/device: 381.489, Tokens/s/device: 57502.246, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0421 12:56:01.537607 137843719460672 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.550, Tokens/s/device: 13347.150, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0421 12:56:01.542490 137843719460672 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0421 12:56:01.832224 137843719460672 checkpointing.py:798] Waited 0.28969764709472656 seconds for step 15 to finish before starting checkpointing. I0421 12:56:01.835238 137843719460672 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 12:56:05.588234 137720005392128 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_7 I0421 12:56:10.846279 137722110912256 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 151.903 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.39873456954956 s) (per-host) I0421 12:56:10.846401 137722110912256 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.111913s. I0421 12:56:17.087602 137722110912256 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 16.353097s. I0421 12:56:17.087939 137719996999424 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 12:56:17.088073 137719996999424 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 12:56:17.088123 137719996999424 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 12:56:17.089693 137719996999424 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 12:56:17.089885 137843719460672 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0421 12:56:17.090019 137843719460672 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.254781 seconds. If this number is high, consider checkpointing less frequently. I0421 12:56:17.092006 137843719460672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 15 I0421 12:56:17.094397 137843719460672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/15. I0421 12:56:17.814392 137843719460672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 12:56:17.814487 137843719460672 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 I0421 12:56:17.842354 137843719460672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029101s I0421 12:56:17.842481 137843719460672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 47.567 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03242921829223633 s) (per-host) I0421 12:56:17.842528 137843719460672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.032485s (batch_requests_ready=0.001743s, total_serialization_initiated=0.030681s, others=0.000060s) I0421 12:56:17.842606 137843719460672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.037525s (all_items=0.000014s, per_item={'items': '0.00001359'}, temp_paths=0.037511) I0421 12:56:17.843248 137843719460672 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/15. I0421 12:56:17.843500 137719996999424 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:16:17.843478 I0421 12:56:17.852892 137843719460672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 12:56:17.853152 137720005392128 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 12:56:17.853276 137843719460672 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776161.835208, 'wait_for_prev_duration_secs': 15.254780769348145, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776776177.0920455, 'checkpointer_blocking_duration_secs': 0.7515518665313721, 'get_old_steps_start_time': 1776776177.8436196, 'get_old_steps_duration_secs': 3.337860107421875e-05, 'checkpoint_manager_blocking_start_time': 1776776161.8328192, 'checkpoint_manager_blocking_duration_secs': 16.020423412322998} I0421 12:56:17.853392 137843719460672 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0421 12:56:17.854104 137843719460672 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.170, Tokens/s/device: 13892.846, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0421 12:56:17.860248 137843719460672 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.127, Tokens/s/device: 13886.440, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0421 12:56:23.373728 137720533870336 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_7 I0421 12:56:28.331860 137719996999424 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 150.126 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.521777391433716 s) (per-host) I0421 12:56:28.331969 137719996999424 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.488394s. I0421 12:56:33.944095 137719996999424 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 16.100505s. I0421 12:56:33.944319 137720005392128 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 12:56:33.944386 137720005392128 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 12:56:33.944433 137720005392128 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 12:56:33.947422 137720005392128 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 12:56:38.590440 137843719460672 metric_logger.py:196] completed step: 16, seconds: 16.316, TFLOP/s/device: 0.833, Tokens/s/device: 125.518, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0421 12:56:38.737918 137843719460672 metric_logger.py:196] completed step: 17, seconds: 20.584, TFLOP/s/device: 0.660, Tokens/s/device: 99.495, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0421 12:56:38.885741 137843719460672 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.984, Tokens/s/device: 13412.709, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0421 12:56:39.032346 137843719460672 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0421 12:56:39.033041 137843719460672 checkpointing.py:798] Waited 0.0007157325744628906 seconds for step 19 to finish before starting checkpointing. I0421 12:56:39.035323 137843719460672 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 12:56:39.037096 137843719460672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 19 I0421 12:56:39.038355 137843719460672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/19. I0421 12:56:39.349669 137843719460672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 12:56:39.349773 137843719460672 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 I0421 12:56:39.376123 137843719460672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027338s I0421 12:56:39.376302 137843719460672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 50.167 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030748605728149414 s) (per-host) I0421 12:56:39.376358 137843719460672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030816s (batch_requests_ready=0.001611s, total_serialization_initiated=0.029133s, others=0.000072s) I0421 12:56:39.376452 137843719460672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034983s (all_items=0.000011s, per_item={'items': '0.00001097'}, temp_paths=0.034972) I0421 12:56:39.377176 137843719460672 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/19. I0421 12:56:39.377487 137720005392128 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:16:39.377451 I0421 12:56:39.379790 137843719460672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 12:56:39.380019 137716230510336 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 12:56:39.380191 137843719460672 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776199.0352933, 'wait_for_prev_duration_secs': 7.128715515136719e-05, 'time_between_consecutive_saves_sec': 5.087831497192383, 'checkpointer_blocking_start_time': 1776776199.0371346, 'checkpointer_blocking_duration_secs': 0.34049224853515625, 'get_old_steps_start_time': 1776776199.3776493, 'get_old_steps_duration_secs': 3.552436828613281e-05, 'checkpoint_manager_blocking_start_time': 1776776199.0332794, 'checkpoint_manager_blocking_duration_secs': 0.3468806743621826} I0421 12:56:39.380311 137843719460672 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0421 12:56:39.380360 137843719460672 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 12:56:44.440185 137720533870336 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_7 I0421 12:56:49.196972 137720005392128 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 160.343 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 9.851387977600098 s) (per-host) I0421 12:56:49.197091 137720005392128 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 9.819494s. I0421 12:56:54.321101 137720005392128 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 14.943482s. I0421 12:56:54.321395 137716230510336 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 12:56:54.321532 137716230510336 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 12:56:54.321576 137716230510336 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 12:56:54.323727 137716230510336 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 12:56:54.323868 137843719460672 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0421 12:56:54.323950 137843719460672 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 12:56:54.324890 137843719460672 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.227, Tokens/s/device: 13901.427, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Tue Apr 21 12:57:06 UTC 2026 EXIT_CODE=0