2026-04-15 21:46:14.723399: 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) I0415 21:46:14.842491 128766452608128 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 21:47:05.353194 128766452608128 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 21:47:05.353305 128766452608128 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 21:47:05.353341 128766452608128 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 I0415 21:47:05.353371 128766452608128 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 21:47:05.353394 128766452608128 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 21:47:05.353469 128766452608128 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 21:47:05.445290 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:05.848753 128766452608128 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 21:47:05.849039 128766452608128 checkpointing.py:677] Setting up checkpoint logger... I0415 21:47:05.849097 128766452608128 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 21:47:05.849139 128766452608128 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 21:47:05.849630 128766452608128 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 0x751c2fffaf60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 21:47:08.237257 128766452608128 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 21:47:08.237427 128766452608128 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 0x7515b4503920>}, handler_registry=None I0415 21:47:08.237723 128766452608128 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7515b4503920>` 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`. I0415 21:47:08.237764 128766452608128 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7515b3d3a0f0>` 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`. I0415 21:47:08.237794 128766452608128 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 0x7515b4503920>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7515b4503920>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7515b3d3a0f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7515b3d3a0f0>}). I0415 21:47:08.238308 128766452608128 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 21:47:08.238362 128766452608128 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7515b3d2d1c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 21:47:08.364867 128766452608128 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints I0415 21:47:08.365116 128766452608128 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_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7515b3d381d0> I0415 21:47:08.365198 128766452608128 checkpointing.py:301] Checkpoint manager created! W0415 21:47:08.381588 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:08.560582 128766452608128 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 21:47:08.560682 128766452608128 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 21:47:08.653985 128766452608128 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 21:47:08.654076 128766452608128 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 21:47:08.668405 128766452608128 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 21:47:08.668452 128766452608128 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 21:47:08.688956 128766452608128 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 21:47:08.689012 128766452608128 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 21:47:08.703423 128766452608128 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 21:47:08.703471 128766452608128 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 21:47:08.717802 128766452608128 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 21:47:08.717850 128766452608128 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 21:47:08.739503 128766452608128 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 21:47:08.739558 128766452608128 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 21:47:08.757797 128766452608128 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 21:47:08.757857 128766452608128 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 21:47:08.939027 128766452608128 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 21:47:08.939130 128766452608128 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0415 21:47:08.990243 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 I0415 21:47:09.968935 128766452608128 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 21:47:09.969068 128766452608128 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 21:47:09.969112 128766452608128 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 21:47:09.969157 128766452608128 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 21:47:09.969196 128766452608128 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 21:47:09.969230 128766452608128 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 21:47:09.969270 128766452608128 maxtext_utils.py:1620] 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) I0415 21:47:09.969308 128766452608128 maxtext_utils.py:1620] 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') I0415 21:47:09.969337 128766452608128 maxtext_utils.py:1620] 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) I0415 21:47:09.969363 128766452608128 maxtext_utils.py:1620] 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) I0415 21:47:09.969395 128766452608128 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0415 21:47:09.969428 128766452608128 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0415 21:47:10.621576 128766452608128 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 21:47:10.621659 128766452608128 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 21:47:10.635210 128766452608128 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 21:47:10.635259 128766452608128 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 21:47:11.391250 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:11.524211 128766452608128 max_utils.py:791] Total memory size: 2.4 GB, Output size: 0.5 GB, Temp size: 1.9 GB, Argument size: 0.5 GB, Host temp size: 0.0 GB. I0415 21:47:13.170399 128766452608128 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 21:47:13.170630 128766452608128 metric_logger.py:289] number parameters: 1.104 billion W0415 21:47:13.979579 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:14.095568 128766452608128 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0415 21:47:14.280894 128766452608128 checkpointing.py:776] Waited 0.1853480339050293 seconds for step 0 to finish before starting checkpointing. I0415 21:47:14.281452 128766452608128 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 21:47:14.281615 128766452608128 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0415 21:47:14.282069 128766452608128 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0. I0415 21:47:14.363028 128766452608128 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 21:47:14.457437 128766452608128 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 21:47:14.457525 128766452608128 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 I0415 21:47:14.463261 128659149030976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0 W0415 21:47:14.672478 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.684169 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.690112 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.698041 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.716984 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.725231 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.729789 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.733820 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:14.903548 128766452608128 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.446460s I0415 21:47:14.904160 128766452608128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 22.853 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 540 milliseconds) (per-host) I0415 21:47:14.904319 128766452608128 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.540338s (batch_requests_ready=0.089470s, total_serialization_initiated=0.450336s, others=0.000531s) I0415 21:47:14.904439 128766452608128 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.540935s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=0.540914) I0415 21:47:14.905282 128659086116416 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 21:47:14.905381 128766452608128 async_checkpointer.py:561] Finished blocking save. Time taken: 0.623725s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0. I0415 21:47:14.905560 128766452608128 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0415 21:47:14.905724 128659163711040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 21:47:14.905837 128766452608128 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776289634.28144, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776289634.2816374, 'checkpointer_blocking_duration_secs': 0.6238467693328857, 'get_old_steps_start_time': 1776289634.9055023, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1776289634.2813454, 'checkpoint_manager_blocking_duration_secs': 0.6244533061981201} I0415 21:47:14.905959 128766452608128 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0415 21:47:14.906022 128766452608128 max_utils.py:750] Memstats: After params initialized: I0415 21:47:14.906126 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_0(process=0,(0,0,0,0)) I0415 21:47:14.906157 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_1(process=0,(1,0,0,0)) I0415 21:47:14.906182 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_2(process=0,(0,1,0,0)) I0415 21:47:14.906208 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_3(process=0,(1,1,0,0)) I0415 21:47:14.906228 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_4(process=0,(0,2,0,0)) I0415 21:47:14.906248 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_5(process=0,(1,2,0,0)) I0415 21:47:14.906269 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_6(process=0,(0,3,0,0)) I0415 21:47:14.906289 128766452608128 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_7(process=0,(1,3,0,0)) W0415 21:47:14.912234 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.917635 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 21:47:14.922231 3264614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 21:47:15.248138 128766452608128 metric_logger.py:185] completed step: 0, seconds: 0.924, TFLOP/s/device: 14.699, Tokens/s/device: 2215.563, total_weights: 16384, loss: 10.874 I0415 21:47:15.248861 128766452608128 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/tensorboard/' I0415 21:47:15.337120 128659138545216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/items I0415 21:47:15.429691 128766452608128 metric_logger.py:185] completed step: 1, seconds: 1.151, TFLOP/s/device: 11.810, Tokens/s/device: 1780.093, total_weights: 16384, loss: 10.874 I0415 21:47:15.481503 128658995938880 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776289635239296770, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776289635239296770, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0415 21:47:15.616430 128766452608128 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 2067.741, Tokens/s/device: 311672.500, total_weights: 16384, loss: 9.775 I0415 21:47:15.802392 128766452608128 metric_logger.py:185] completed step: 3, seconds: 0.183, TFLOP/s/device: 74.100, Tokens/s/device: 11169.102, total_weights: 16384, loss: 8.838 I0415 21:47:15.838024 3266781 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 21:47:15.838091 3266781 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 21:47:16.623017 128659107087936 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_0 I0415 21:47:35.676967 128766452608128 metric_logger.py:185] completed step: 4, seconds: 0.186, TFLOP/s/device: 72.969, Tokens/s/device: 10998.690, total_weights: 16384, loss: 7.965 I0415 21:47:35.683964 128766452608128 metric_logger.py:185] completed step: 5, seconds: 0.186, TFLOP/s/device: 72.997, Tokens/s/device: 11002.944, total_weights: 16384, loss: 7.233 I0415 21:47:35.866227 128766452608128 metric_logger.py:185] completed step: 6, seconds: 19.874, TFLOP/s/device: 0.684, Tokens/s/device: 103.047, total_weights: 16384, loss: 6.677 I0415 21:47:36.053822 128766452608128 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 2933.951, Tokens/s/device: 442237.098, total_weights: 16384, loss: 6.293 I0415 21:47:36.241384 128766452608128 metric_logger.py:185] completed step: 8, seconds: 0.183, TFLOP/s/device: 74.133, Tokens/s/device: 11174.099, total_weights: 16384, loss: 6.055 I0415 21:47:36.427950 128766452608128 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0415 21:47:36.431425 128766452608128 checkpointing.py:776] Waited 0.0035028457641601562 seconds for step 9 to finish before starting checkpointing. I0415 21:47:36.431693 128766452608128 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 21:47:44.649743 128659096602176 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 28.809904s (commit=28.388870s, array_metadata_write=0.421034s) I0415 21:47:44.650793 128659086116416 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 417.238 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 30 seconds) (per-host) I0415 21:47:44.650895 128659086116416 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.745468s. I0415 21:47:44.908023 128659086116416 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776289635239296770, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0415 21:47:45.078562 128659086116416 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 21:47:45.299285 128658995938880 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776289635239296770, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0415 21:47:45.517542 128659086116416 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 21:47:45.518175 128659086116416 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.566059s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/items I0415 21:47:45.518817 128659086116416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0/items I0415 21:47:45.747310 128659086116416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0 I0415 21:47:46.451154 128659086116416 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0`. I0415 21:47:46.451784 128659086116416 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 32.170135s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/0 I0415 21:47:46.451856 128659086116416 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.546432s. I0415 21:47:46.452066 128659163711040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 21:47:46.452177 128659163711040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 21:47:46.452233 128659163711040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0415 21:47:46.452281 128659163711040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0415 21:47:46.452422 128766452608128 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0415 21:47:46.452549 128766452608128 checkpoint_manager.py:1441] Waiting for previous save to complete took 10.020859 seconds. If this number is high, consider checkpointing less frequently. I0415 21:47:46.453308 128766452608128 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0415 21:47:46.453588 128766452608128 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9. I0415 21:47:46.645964 128766452608128 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 21:47:46.646106 128659163711040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9 I0415 21:47:46.646183 128766452608128 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 I0415 21:47:46.834039 128766452608128 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.188838s I0415 21:47:46.834479 128766452608128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 43.262 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 285 milliseconds) (per-host) I0415 21:47:46.834534 128766452608128 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285349s (batch_requests_ready=0.093359s, total_serialization_initiated=0.191656s, others=0.000334s) I0415 21:47:46.834595 128766452608128 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285849s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.285833) I0415 21:47:46.835355 128659046270528 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 21:47:46.835453 128766452608128 async_checkpointer.py:561] Finished blocking save. Time taken: 0.382099s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9. I0415 21:47:46.835634 128766452608128 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0415 21:47:46.835809 128659035784768 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 21:47:46.835932 128766452608128 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776289656.4316587, 'wait_for_prev_duration_secs': 10.020859241485596, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776289666.4533327, 'checkpointer_blocking_duration_secs': 0.3822181224822998, 'get_old_steps_start_time': 1776289666.8355658, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776289656.431619, 'checkpoint_manager_blocking_duration_secs': 10.404274940490723} I0415 21:47:46.836055 128766452608128 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0415 21:47:46.836090 128766452608128 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 21:47:47.365135 128658943510080 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9/items I0415 21:47:48.613927 128659067242048 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_0 I0415 21:48:22.552120 128659056756288 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.689719s (commit=34.265871s, array_metadata_write=0.423849s) I0415 21:48:22.553339 128659046270528 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 350.982 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0415 21:48:22.553452 128659046270528 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.717952s. I0415 21:48:22.990298 128659046270528 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 21:48:23.456486 128659046270528 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 21:48:23.456676 128659046270528 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.614796s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9/items I0415 21:48:23.457319 128659046270528 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9/items I0415 21:48:23.723027 128659046270528 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9 I0415 21:48:24.401709 128659046270528 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9`. I0415 21:48:24.402327 128659046270528 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 37.948978s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_11_optimizer_offload_true/checkpoints/9 I0415 21:48:24.402408 128659046270528 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.566910s. I0415 21:48:24.402585 128659035784768 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 21:48:24.402694 128659035784768 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 21:48:24.402768 128659035784768 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0415 21:48:24.402813 128659035784768 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0415 21:48:24.402963 128766452608128 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. I0415 21:48:24.403110 128766452608128 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 21:48:24.404003 128766452608128 metric_logger.py:185] completed step: 9, seconds: 0.187, TFLOP/s/device: 72.467, Tokens/s/device: 10922.958, total_weights: 16384, loss: 5.915 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops