XPK Start: Sun Apr 19 02:37:55 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 02:38:20.123470: 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) I0419 02:38:20.302745 133096389539648 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 02:38:29.343804 133096389539648 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 02:38:29.346123 133096389539648 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-2h094-slice-job-0-0.mt-16-zarr3-true-sv-2h094:8482 I0419 02:38:30.555388 133096389539648 max_utils.py:284] Jax distributed system initialized! I0419 02:38:36.428002 133096389539648 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 02:38:36.428107 133096389539648 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 02:38:36.428147 133096389539648 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 I0419 02:38:36.428182 133096389539648 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 02:38:37.125164 133096389539648 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 02:38:37.238128 133096389539648 checkpointing.py:688] Setting up checkpoint logger... I0419 02:38:37.238253 133096389539648 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 02:38:37.238301 133096389539648 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 02:38:37.238502 133096389539648 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 0x790c3393fb30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 02:38:41.048317 133096389539648 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 02:38:41.048615 133096389539648 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 0x78f8b432ab10>}, handler_registry=None I0419 02:38:41.048856 133096389539648 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f8b432ab10>` 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`. I0419 02:38:41.048904 133096389539648 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f8b435c080>` 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`. I0419 02:38:41.048939 133096389539648 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 0x78f8b432ab10>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f8b432ab10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f8b435c080>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f8b435c080>}). I0419 02:38:41.049263 133096389539648 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 02:38:41.049337 133096389539648 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78f8b43e8fe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 02:38:42.590262 133096389539648 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints I0419 02:38:43.043205 133096389539648 checkpoint_manager.py:929] [process=6][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78f8cc380fe0> I0419 02:38:43.043423 133096389539648 checkpointing.py:302] Checkpoint manager created! I0419 02:38:44.002440 133096389539648 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 02:38:44.002562 133096389539648 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 02:38:45.382708 133096389539648 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 02:38:45.382824 133096389539648 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:38:45.382876 133096389539648 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:38:45.382920 133096389539648 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 02:38:45.382956 133096389539648 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:38:45.382992 133096389539648 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:38:45.383028 133096389539648 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:38:45.383063 133096389539648 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 02:38:45.383096 133096389539648 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:38:45.383129 133096389539648 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:38:45.383161 133096389539648 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 02:38:45.383192 133096389539648 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 02:38:45.778550 133096389539648 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:38:45.778645 133096389539648 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:38:45.783945 133096389539648 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:38:45.784000 133096389539648 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:38:45.800408 133096389539648 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:38:45.800467 133096389539648 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:38:45.816187 133096389539648 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:38:45.816257 133096389539648 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:38:45.841439 133096389539648 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:38:45.841508 133096389539648 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:38:45.857338 133096389539648 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:38:45.857403 133096389539648 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:38:45.873180 133096389539648 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:38:45.873259 133096389539648 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:38:45.903162 133096389539648 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 02:38:45.903242 133096389539648 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:38:45.922875 133096389539648 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 02:38:45.922936 133096389539648 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 02:38:56.977276 133096389539648 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. I0419 02:38:56.979660 133096389539648 metric_logger.py:301] number parameters: 1.104 billion I0419 02:39:08.311533 133096389539648 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 02:39:08.464472 133096389539648 checkpointing.py:798] Waited 0.1529245376586914 seconds for step 0 to finish before starting checkpointing. I0419 02:39:08.467110 133096389539648 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:39:08.468819 133096389539648 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0419 02:39:08.470352 133096389539648 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/0. I0419 02:39:09.205868 133096389539648 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 02:39:09.206929 133096389539648 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:39:09.206987 133096389539648 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 I0419 02:39:09.484402 133096389539648 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.279307s I0419 02:39:09.484590 133096389539648 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.350 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2883477210998535 s) (per-host) I0419 02:39:09.484654 133096389539648 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288423s (batch_requests_ready=0.002925s, total_serialization_initiated=0.285412s, others=0.000086s) I0419 02:39:09.484786 133096389539648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292595s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.292577) I0419 02:39:09.485631 133096389539648 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/0. I0419 02:39:09.485919 132971163289344 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:59:09.485884 I0419 02:39:09.491428 133096389539648 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:39:09.491734 132965782775552 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:39:09.491893 133096389539648 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776566348.4670923, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776566348.4688554, 'checkpointer_blocking_duration_secs': 1.0172185897827148, 'get_old_steps_start_time': 1776566349.4861, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776566348.4648697, 'checkpoint_manager_blocking_duration_secs': 1.026984453201294} I0419 02:39:09.492006 133096389539648 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 02:39:09.492072 133096389539648 max_utils.py:750] Memstats: After params initialized: I0419 02:39:09.492121 133096389539648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0419 02:39:09.492153 133096389539648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0419 02:39:09.492178 133096389539648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0419 02:39:09.492200 133096389539648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0419 02:39:09.805094 133096389539648 metric_logger.py:196] completed step: 0, seconds: 11.331, TFLOP/s/device: 1.199, Tokens/s/device: 180.747, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:39:09.979514 133096389539648 metric_logger.py:196] completed step: 1, seconds: 1.492, TFLOP/s/device: 9.108, Tokens/s/device: 1372.830, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:39:10.413227 133096389539648 metric_logger.py:196] completed step: 2, seconds: 0.028, TFLOP/s/device: 482.189, Tokens/s/device: 72680.815, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0419 02:39:10.560746 133096389539648 metric_logger.py:196] completed step: 3, seconds: 0.424, TFLOP/s/device: 32.081, Tokens/s/device: 4835.537, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0419 02:39:10.566442 133096389539648 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0419 02:39:10.856001 133096389539648 checkpointing.py:798] Waited 0.2895324230194092 seconds for step 5 to finish before starting checkpointing. I0419 02:39:10.858902 133096389539648 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:39:13.080071 2787 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 02:39:15.083055 132968442029824 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/0/items/array_metadatas/process_6 I0419 02:39:44.789782 132971163289344 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 44.379 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 35.59352135658264 s) (per-host) I0419 02:39:44.789886 132971163289344 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.303853s. I0419 02:39:56.267178 132971163289344 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 46.781126s. I0419 02:39:56.267488 132965782775552 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:39:56.267639 132965782775552 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:39:56.267705 132965782775552 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:39:56.269833 132965782775552 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 02:39:56.270029 133096389539648 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 02:39:56.270185 133096389539648 checkpoint_manager.py:1452] Waiting for previous save to complete took 45.411275 seconds. If this number is high, consider checkpointing less frequently. I0419 02:39:56.272375 133096389539648 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0419 02:39:56.274783 133096389539648 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/5. I0419 02:39:56.991049 133096389539648 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:39:56.991147 133096389539648 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 I0419 02:39:57.035972 133096389539648 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046582s I0419 02:39:57.036135 133096389539648 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 28.861 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05344796180725098 s) (per-host) I0419 02:39:57.036183 133096389539648 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.053505s (batch_requests_ready=0.002548s, total_serialization_initiated=0.050894s, others=0.000064s) I0419 02:39:57.036286 133096389539648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.058511s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.058496) I0419 02:39:57.036990 133096389539648 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/5. I0419 02:39:57.037374 132965782775552 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:59:57.037337 I0419 02:39:57.045912 133096389539648 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:39:57.046185 132942785406720 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:39:57.046373 133096389539648 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776566350.858874, 'wait_for_prev_duration_secs': 45.411274671554565, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776566396.2724142, 'checkpointer_blocking_duration_secs': 0.7651107311248779, 'get_old_steps_start_time': 1776566397.0375516, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776566350.8563735, 'checkpoint_manager_blocking_duration_secs': 46.1899631023407} I0419 02:39:57.046490 133096389539648 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0419 02:39:57.047566 133096389539648 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.716, Tokens/s/device: 12467.811, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0419 02:39:57.055416 133096389539648 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.361, Tokens/s/device: 13921.650, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0419 02:40:02.954604 132968442029824 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/5/items/array_metadatas/process_6 I0419 02:40:21.016122 133096389539648 metric_logger.py:196] completed step: 6, seconds: 46.487, TFLOP/s/device: 0.292, Tokens/s/device: 44.055, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0419 02:40:21.163777 133096389539648 metric_logger.py:196] completed step: 7, seconds: 23.815, TFLOP/s/device: 0.571, Tokens/s/device: 85.996, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0419 02:40:21.311475 133096389539648 metric_logger.py:196] completed step: 8, seconds: 0.154, TFLOP/s/device: 88.397, Tokens/s/device: 13324.138, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0419 02:40:21.459948 133096389539648 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 02:40:21.460659 133096389539648 checkpointing.py:798] Waited 0.0007245540618896484 seconds for step 9 to finish before starting checkpointing. I0419 02:40:21.462616 133096389539648 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:40:40.323607 132965782775552 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.446 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.34088063240051 s) (per-host) I0419 02:40:40.323726 132965782775552 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.286237s. I0419 02:40:46.983789 132965782775552 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.946287s. I0419 02:40:46.984090 132942785406720 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:40:46.984209 132942785406720 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:40:46.984276 132942785406720 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:40:46.987121 132942785406720 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0419 02:40:46.987328 133096389539648 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0419 02:40:46.987453 133096389539648 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.524844 seconds. If this number is high, consider checkpointing less frequently. I0419 02:40:46.988954 133096389539648 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0419 02:40:46.990865 133096389539648 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/9. I0419 02:40:47.693439 133096389539648 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:40:47.693535 133096389539648 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 I0419 02:40:47.734157 133096389539648 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042391s I0419 02:40:47.734330 133096389539648 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.359 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.049190521240234375 s) (per-host) I0419 02:40:47.734382 133096389539648 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049254s (batch_requests_ready=0.002639s, total_serialization_initiated=0.046546s, others=0.000068s) I0419 02:40:47.734472 133096389539648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053451s (all_items=0.000010s, per_item={'items': '0.00001025'}, temp_paths=0.053441) I0419 02:40:47.735108 133096389539648 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/9. I0419 02:40:47.735449 132942785406720 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 03:00:47.735414 I0419 02:40:47.737336 133096389539648 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:40:47.737600 132971138111232 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:40:47.737752 133096389539648 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776566421.4625857, 'wait_for_prev_duration_secs': 25.5248441696167, 'time_between_consecutive_saves_sec': 25.192713260650635, 'checkpointer_blocking_start_time': 1776566446.988997, 'checkpointer_blocking_duration_secs': 0.7465593814849854, 'get_old_steps_start_time': 1776566447.7355728, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776566421.4609, 'checkpoint_manager_blocking_duration_secs': 26.276822090148926} I0419 02:40:47.737867 133096389539648 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 02:40:47.737910 133096389539648 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:40:53.798604 132968442029824 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_true/checkpoints/9/items/array_metadatas/process_6 I0419 02:41:29.760411 132942785406720 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.542 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.07525110244751 s) (per-host) I0419 02:41:29.760518 132942785406720 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.024456s. I0419 02:41:37.834459 132942785406720 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.098380s. I0419 02:41:37.834773 132971138111232 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:41:37.834893 132971138111232 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:41:37.834937 132971138111232 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:41:37.837139 132971138111232 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 02:41:37.837353 133096389539648 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 02:41:37.837539 133096389539648 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:41:37.838520 133096389539648 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.817, Tokens/s/device: 13839.615, 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: Sun Apr 19 02:41:46 UTC 2026 EXIT_CODE=0