2026-04-17 13:00:59.772500: 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 13:00:59.887982 131496625142912 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:01:34.720798 131496625142912 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:01:34.720915 131496625142912 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:01:34.720951 131496625142912 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0417 13:01:34.720980 131496625142912 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:01:34.721204 131496625142912 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:01:34.814273 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:01:35.212760 131496625142912 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:01:35.213375 131496625142912 checkpointing.py:677] Setting up checkpoint logger... I0417 13:01:35.213428 131496625142912 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:01:35.213492 131496625142912 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:01:35.214162 131496625142912 base_pytree_checkpoint_handler.py:411] 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 0x7797dabf8a10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:01:37.673944 131496625142912 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:01:37.674154 131496625142912 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x779183b28470>}, handler_registry=None I0417 13:01:37.674469 131496625142912 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x779183b28470>` 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 13:01:37.674513 131496625142912 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x779182dd2630>` 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 13:01:37.674542 131496625142912 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 0x779183b28470>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x779183b28470>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x779182dd2630>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x779182dd2630>}). I0417 13:01:37.674846 131496625142912 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:01:37.674901 131496625142912 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x77916012de40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:01:37.804114 131496625142912 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints I0417 13:01:37.804352 131496625142912 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7792dc8b0d40> I0417 13:01:37.804437 131496625142912 checkpointing.py:301] Checkpoint manager created! I0417 13:01:38.237284 131496625142912 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:01:38.240161 131496625142912 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0417 13:01:38.291990 131496625142912 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:01:38.315500 131496625142912 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:01:38.315562 131496625142912 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:01:39.095448 131496625142912 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:01:39.095551 131496625142912 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:01:39.198616 131496625142912 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:01:39.198695 131496625142912 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:01:39.213615 131496625142912 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:01:39.213662 131496625142912 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:01:39.269837 131496625142912 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:01:39.269907 131496625142912 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:01:39.284859 131496625142912 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:01:39.284913 131496625142912 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:01:39.299813 131496625142912 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:01:39.299863 131496625142912 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:01:39.321958 131496625142912 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:01:39.322025 131496625142912 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:01:39.378358 131496625142912 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:01:39.378425 131496625142912 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:01:39.818058 131496625142912 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:01:39.818152 131496625142912 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0417 13:01:42.073184 131496625142912 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:01:42.073291 131496625142912 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:01:42.073332 131496625142912 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:01:42.073377 131496625142912 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 13:01:42.073418 131496625142912 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:01:42.073447 131496625142912 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:01:42.073487 131496625142912 maxtext_utils.py:1651] 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 13:01:42.073526 131496625142912 maxtext_utils.py:1651] 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 13:01:42.073554 131496625142912 maxtext_utils.py:1651] 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 13:01:42.073579 131496625142912 maxtext_utils.py:1651] 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 13:01:42.073614 131496625142912 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:01:42.073649 131496625142912 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:01:43.263564 131496625142912 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:01:43.263746 131496625142912 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:01:43.278011 131496625142912 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:01:43.278073 131496625142912 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:01:58.917215 131496625142912 max_utils.py:791] Total memory size: 3.7 GB, Output size: 1.5 GB, Temp size: 2.2 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0417 13:01:58.917890 131496625142912 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:01:58.918068 131496625142912 metric_logger.py:301] number parameters: 1.104 billion I0417 13:02:17.096170 131496625142912 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:02:17.240298 131496625142912 checkpointing.py:776] Waited 0.14411401748657227 seconds for step 0 to finish before starting checkpointing. I0417 13:02:17.240830 131496625142912 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:02:17.241018 131496625142912 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:02:17.241418 131496625142912 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0. I0417 13:02:17.329657 131496625142912 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:02:17.413752 131380971308608 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0 I0417 13:02:17.430194 131496625142912 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:02:17.430289 131496625142912 replica_slices.py:410] 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 W0417 13:02:18.050313 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.062311 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.076654 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.083748 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.089070 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.093846 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.098657 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.103531 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:02:18.430184 131380960822848 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/items I0417 13:02:18.580595 131380920976960 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776430938338358720, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776430938338358720, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0417 13:02:18.600891 131496625142912 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.171072s I0417 13:02:18.601542 131496625142912 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 9.713 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0417 13:02:18.601667 131496625142912 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.270757s (batch_requests_ready=0.094859s, total_serialization_initiated=1.175369s, others=0.000529s) I0417 13:02:18.601789 131496625142912 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.271662s (all_items=0.000024s, per_item={'items': '0.00002432'}, temp_paths=1.271638) I0417 13:02:18.602855 131380900005440 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:02:18.602957 131496625142912 async_checkpointer.py:561] Finished blocking save. Time taken: 1.361882s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0. I0417 13:02:18.603158 131496625142912 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:02:18.603375 131380889519680 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:02:18.603480 131496625142912 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776430937.2408168, 'wait_for_prev_duration_secs': 4.315376281738281e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776430937.241057, 'checkpointer_blocking_duration_secs': 1.3620238304138184, 'get_old_steps_start_time': 1776430938.6031017, 'get_old_steps_duration_secs': 1.9311904907226562e-05, 'checkpoint_manager_blocking_start_time': 1776430937.2407162, 'checkpoint_manager_blocking_duration_secs': 1.3627355098724365} I0417 13:02:18.603604 131496625142912 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:02:18.603645 131496625142912 max_utils.py:750] Memstats: After params initialized: I0417 13:02:18.603745 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:02:18.603774 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:02:18.603795 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:02:18.603814 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:02:18.603833 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:02:18.603851 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:02:18.603868 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:02:18.603886 131496625142912 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:02:18.610647 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.616977 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:02:18.622447 419097 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:02:18.933367 424393 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:02:18.933424 424393 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:02:18.964972 131496625142912 metric_logger.py:196] completed step: 0, seconds: 18.177, TFLOP/s/device: 0.747, Tokens/s/device: 112.667, total_weights: 13328, loss: 10.840, lm_loss: 10.840, perplexity: 51025.957 I0417 13:02:18.965934 131496625142912 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/tensorboard/' I0417 13:02:19.068160 131496625142912 metric_logger.py:196] completed step: 1, seconds: 1.850, TFLOP/s/device: 7.343, Tokens/s/device: 1106.757, total_weights: 12332, loss: 10.869, lm_loss: 10.869, perplexity: 52515.969 I0417 13:02:19.190187 131496625142912 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 505.417, Tokens/s/device: 76181.974, total_weights: 15161, loss: 9.874, lm_loss: 9.874, perplexity: 19421.770 I0417 13:02:19.312544 131496625142912 metric_logger.py:196] completed step: 3, seconds: 0.104, TFLOP/s/device: 130.451, Tokens/s/device: 19663.002, total_weights: 13327, loss: 9.406, lm_loss: 9.406, perplexity: 12166.079 I0417 13:02:19.710733 131380950337088 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/items/array_metadatas/process_0 I0417 13:02:43.191356 131380971308608 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.264636s (commit=23.831392s, array_metadata_write=0.433244s) I0417 13:02:43.192663 131380900005440 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 488.630 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host) I0417 13:02:43.192786 131380900005440 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.589781s. I0417 13:02:43.409490 131380900005440 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776430938338358720, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0417 13:02:43.596517 131380900005440 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:02:43.796635 131380920976960 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776430938338358720, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0417 13:02:44.032601 131380900005440 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:02:44.033295 131380900005440 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.577965s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/items I0417 13:02:44.033957 131380900005440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0/items I0417 13:02:44.282185 131380900005440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0 I0417 13:02:44.964513 131380900005440 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0`. I0417 13:02:44.965115 131380900005440 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 27.724046s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/0 I0417 13:02:44.965196 131380900005440 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 26.362194s. I0417 13:02:44.965329 131380889519680 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:02:44.965382 131380889519680 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:02:44.965448 131380889519680 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:02:44.965493 131380889519680 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:02:48.160313 131496625142912 metric_logger.py:196] completed step: 4, seconds: 0.122, TFLOP/s/device: 111.438, Tokens/s/device: 16797.211, total_weights: 11939, loss: 9.149, lm_loss: 9.149, perplexity: 9403.998 I0417 13:02:48.170150 131496625142912 metric_logger.py:196] completed step: 5, seconds: 0.122, TFLOP/s/device: 110.953, Tokens/s/device: 16724.101, total_weights: 15502, loss: 9.058, lm_loss: 9.058, perplexity: 8587.564 I0417 13:02:48.290733 131496625142912 metric_logger.py:196] completed step: 6, seconds: 28.847, TFLOP/s/device: 0.471, Tokens/s/device: 70.995, total_weights: 13864, loss: 8.892, lm_loss: 8.892, perplexity: 7274.692 I0417 13:02:48.413288 131496625142912 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2061.780, Tokens/s/device: 310773.900, total_weights: 12988, loss: 8.751, lm_loss: 8.751, perplexity: 6314.503 I0417 13:02:48.535499 131496625142912 metric_logger.py:196] completed step: 8, seconds: 0.122, TFLOP/s/device: 111.629, Tokens/s/device: 16825.916, total_weights: 13820, loss: 8.791, lm_loss: 8.791, perplexity: 6573.798 I0417 13:02:48.657375 131496625142912 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:02:48.658264 131496625142912 checkpointing.py:776] Waited 0.0009093284606933594 seconds for step 9 to finish before starting checkpointing. I0417 13:02:48.658527 131496625142912 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:02:48.658694 131496625142912 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:02:48.658977 131496625142912 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9. I0417 13:02:48.831908 131380889519680 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9 I0417 13:02:48.836593 131496625142912 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:02:48.836688 131496625142912 replica_slices.py:410] 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 13:02:49.527354 131380960822848 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9/items I0417 13:02:56.869822 131496625142912 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.033612s I0417 13:02:56.875414 131496625142912 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.518 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0417 13:02:56.875495 131496625142912 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.130239s (batch_requests_ready=0.087869s, total_serialization_initiated=8.036958s, others=0.005412s) I0417 13:02:56.875591 131496625142912 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.130803s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=8.130784) I0417 13:02:56.877719 131380889519680 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:02:56.877861 131496625142912 async_checkpointer.py:561] Finished blocking save. Time taken: 8.219118s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9. I0417 13:02:56.878130 131496625142912 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:02:56.878419 131380900005440 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:02:56.878597 131496625142912 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776430968.658493, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': 3.6929850578308105, 'checkpointer_blocking_start_time': 1776430968.658716, 'checkpointer_blocking_duration_secs': 8.219288110733032, 'get_old_steps_start_time': 1776430976.878028, 'get_old_steps_duration_secs': 5.054473876953125e-05, 'checkpoint_manager_blocking_start_time': 1776430968.658459, 'checkpoint_manager_blocking_duration_secs': 8.22010087966919} I0417 13:02:56.878890 131496625142912 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:02:56.878932 131496625142912 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:02:57.680148 131380950337088 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9/items/array_metadatas/process_0 I0417 13:03:34.455924 131380960822848 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.579552s (commit=37.095683s, array_metadata_write=0.483869s) I0417 13:03:34.457451 131380889519680 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 276.443 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0417 13:03:34.457560 131380889519680 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.579636s. I0417 13:03:34.912648 131380889519680 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:03:35.381858 131380889519680 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:03:35.382969 131380889519680 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.627816s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9/items I0417 13:03:35.383863 131380889519680 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9/items I0417 13:03:35.621529 131380889519680 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9 I0417 13:03:36.307672 131380889519680 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9`. I0417 13:03:36.308593 131380889519680 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.649862s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_04_int8/checkpoints/9 I0417 13:03:36.308707 131380889519680 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.430788s. I0417 13:03:36.308905 131380900005440 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:03:36.309065 131380900005440 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:03:36.309195 131380900005440 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:03:36.309290 131380900005440 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:03:36.309501 131496625142912 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 13:03:36.309699 131496625142912 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:03:36.310701 131496625142912 metric_logger.py:196] completed step: 9, seconds: 0.122, TFLOP/s/device: 110.949, Tokens/s/device: 16723.419, total_weights: 12300, loss: 8.774, lm_loss: 8.774, perplexity: 6465.313 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops