2026-04-16 12:50:09.929097: 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) I0416 12:50:10.049151 133686131514496 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 12:51:05.446664 133686131514496 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 12:51:05.446769 133686131514496 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 12:51:05.446808 133686131514496 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 I0416 12:51:05.446833 133686131514496 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0416 12:51:05.446855 133686131514496 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 12:51:05.446942 133686131514496 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 12:51:05.540713 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 12:51:05.937852 133686131514496 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 12:51:05.938182 133686131514496 checkpointing.py:677] Setting up checkpoint logger... I0416 12:51:05.938228 133686131514496 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 12:51:05.938269 133686131514496 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 12:51:05.938866 133686131514496 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 0x7995a41dafc0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 12:51:08.374817 133686131514496 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0416 12:51:08.374989 133686131514496 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 0x798f2793cda0>}, handler_registry=None I0416 12:51:08.375267 133686131514496 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x798f2793cda0>` 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`. I0416 12:51:08.375308 133686131514496 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x798f2879c140>` 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`. I0416 12:51:08.375338 133686131514496 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 0x798f2793cda0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x798f2793cda0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x798f2879c140>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x798f2879c140>}). I0416 12:51:08.375633 133686131514496 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 12:51:08.375684 133686131514496 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x798f2792d300> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 12:51:08.514809 133686131514496 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints I0416 12:51:08.515061 133686131514496 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_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x798f2793c500> I0416 12:51:08.515145 133686131514496 checkpointing.py:301] Checkpoint manager created! I0416 12:51:09.365109 133686131514496 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 12:51:09.365206 133686131514496 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 12:51:09.463171 133686131514496 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 12:51:09.463249 133686131514496 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 12:51:09.477465 133686131514496 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 12:51:09.477514 133686131514496 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 12:51:09.498871 133686131514496 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 12:51:09.498924 133686131514496 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 12:51:09.512978 133686131514496 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 12:51:09.513026 133686131514496 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 12:51:09.527924 133686131514496 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 12:51:09.528029 133686131514496 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 12:51:09.550114 133686131514496 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 12:51:09.550171 133686131514496 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 12:51:09.568480 133686131514496 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 12:51:09.568537 133686131514496 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 12:51:09.752294 133686131514496 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0416 12:51:09.752413 133686131514496 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0416 12:51:09.803893 3834614 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 I0416 12:51:10.184770 133686131514496 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 12:51:10.184875 133686131514496 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) I0416 12:51:10.184915 133686131514496 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) I0416 12:51:10.184957 133686131514496 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') I0416 12:51:10.184995 133686131514496 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) I0416 12:51:10.185022 133686131514496 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) I0416 12:51:10.185079 133686131514496 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) I0416 12:51:10.185123 133686131514496 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') I0416 12:51:10.185152 133686131514496 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) I0416 12:51:10.185177 133686131514496 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) I0416 12:51:10.185212 133686131514496 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 12:51:10.185245 133686131514496 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 12:51:10.203421 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0416 12:51:10.203471 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0416 12:51:10.216294 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0416 12:51:10.216339 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0416 12:51:10.241882 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0416 12:51:10.241937 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0416 12:51:10.254879 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0416 12:51:10.254925 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0416 12:51:10.280415 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0416 12:51:10.280465 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0416 12:51:10.293278 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0416 12:51:10.293323 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0416 12:51:10.331516 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0416 12:51:10.331576 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0416 12:51:10.344380 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0416 12:51:10.344426 133686131514496 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0416 12:51:10.923097 133686131514496 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 12:51:10.923176 133686131514496 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 12:51:10.936702 133686131514496 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 12:51:10.936751 133686131514496 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 12:51:11.761364 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 12:51:11.892658 133686131514496 max_utils.py:791] Total memory size: 4.5 GB, Output size: 1.5 GB, Temp size: 2.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0416 12:51:11.893113 133686131514496 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 12:51:11.893252 133686131514496 metric_logger.py:289] number parameters: 1.104 billion W0416 12:51:13.243882 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 12:51:13.345633 133686131514496 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0416 12:51:13.767534 133686131514496 checkpointing.py:776] Waited 0.42188572883605957 seconds for step 0 to finish before starting checkpointing. I0416 12:51:13.767959 133686131514496 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 12:51:13.768139 133686131514496 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 12:51:13.768490 133686131514496 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0. I0416 12:51:13.864903 133686131514496 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 12:51:13.953369 133578637706816 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0 I0416 12:51:13.959174 133686131514496 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 12:51:13.959265 133686131514496 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 W0416 12:51:14.186475 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.201318 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.223340 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.239676 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.246650 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.252986 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.257245 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.261431 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 12:51:14.700550 133578627221056 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/items I0416 12:51:14.703546 133686131514496 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.744767s I0416 12:51:14.703925 133686131514496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 14.731 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 837 milliseconds) (per-host) I0416 12:51:14.703979 133686131514496 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.837816s (batch_requests_ready=0.088948s, total_serialization_initiated=0.748587s, others=0.000280s) I0416 12:51:14.704041 133686131514496 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.838656s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.838637) I0416 12:51:14.704815 133578570597952 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 12:51:14.704914 133686131514496 async_checkpointer.py:561] Finished blocking save. Time taken: 0.936731s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0. I0416 12:51:14.705113 133686131514496 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 12:51:14.705307 133686131514496 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776343873.7679453, 'wait_for_prev_duration_secs': 4.172325134277344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776343873.7681623, 'checkpointer_blocking_duration_secs': 0.9368650913238525, 'get_old_steps_start_time': 1776343874.7050576, 'get_old_steps_duration_secs': 2.1696090698242188e-05, 'checkpoint_manager_blocking_start_time': 1776343873.7678654, 'checkpoint_manager_blocking_duration_secs': 0.9374179840087891} I0416 12:51:14.705376 133686131514496 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0416 12:51:14.705414 133686131514496 max_utils.py:750] Memstats: After params initialized: I0416 12:51:14.705489 133686131514496 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 12:51:14.705516 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 12:51:14.705537 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 12:51:14.705555 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 12:51:14.705572 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 12:51:14.705589 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 12:51:14.705606 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 12:51:14.705622 133686131514496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 12:51:14.711258 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.720427 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 12:51:14.725508 3834614 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 12:51:14.854021 133578591569472 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343874612740878, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776343874612740878, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0416 12:51:14.854313 133578648192576 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 12:51:15.047167 133686131514496 metric_logger.py:185] completed step: 0, seconds: 1.452, TFLOP/s/device: 37.432, Tokens/s/device: 5642.141, total_weights: 65536, loss: 10.871 I0416 12:51:15.048158 133686131514496 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/tensorboard/' I0416 12:51:15.380353 3836830 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 12:51:15.380410 3836830 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 12:51:15.468387 133686131514496 metric_logger.py:185] completed step: 1, seconds: 1.700, TFLOP/s/device: 31.977, Tokens/s/device: 4819.898, total_weights: 65536, loss: 10.871 I0416 12:51:15.892531 133686131514496 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 8276.004, Tokens/s/device: 1247449.368, total_weights: 65536, loss: 10.261 I0416 12:51:16.175099 133578637706816 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/items/array_metadatas/process_0 I0416 12:51:16.316213 133686131514496 metric_logger.py:185] completed step: 3, seconds: 0.422, TFLOP/s/device: 128.752, Tokens/s/device: 19406.942, total_weights: 65536, loss: 9.732 I0416 12:51:44.883131 133578616735296 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.507247s (commit=29.043446s, array_metadata_write=0.463801s) I0416 12:51:44.884364 133578570597952 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 407.400 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 31 seconds) (per-host) I0416 12:51:44.884480 133578570597952 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.179522s. I0416 12:51:45.121335 133578570597952 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343874612740878, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0416 12:51:45.317516 133578570597952 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 12:51:45.505102 133578591569472 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343874612740878, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0416 12:51:45.715978 133578570597952 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 12:51:45.716603 133578570597952 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557421s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/items I0416 12:51:45.717273 133578570597952 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0/items I0416 12:51:45.946276 133578570597952 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0 I0416 12:51:46.626597 133578570597952 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0`. I0416 12:51:46.627230 133578570597952 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 32.859056s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/0 I0416 12:51:46.627305 133578570597952 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.922348s. I0416 12:51:46.627513 133578648192576 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 12:51:46.627620 133578648192576 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 12:51:46.627696 133578648192576 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 12:51:46.627746 133578648192576 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 12:51:53.475684 133686131514496 metric_logger.py:185] completed step: 4, seconds: 0.424, TFLOP/s/device: 128.198, Tokens/s/device: 19323.398, total_weights: 65536, loss: 9.274 I0416 12:51:53.482767 133686131514496 metric_logger.py:185] completed step: 5, seconds: 0.424, TFLOP/s/device: 128.231, Tokens/s/device: 19328.368, total_weights: 65536, loss: 8.888 I0416 12:51:53.909800 133686131514496 metric_logger.py:185] completed step: 6, seconds: 37.160, TFLOP/s/device: 1.463, Tokens/s/device: 220.452, total_weights: 65536, loss: 8.588 I0416 12:51:54.344036 133686131514496 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 11822.605, Tokens/s/device: 1782031.760, total_weights: 65536, loss: 8.380 I0416 12:51:54.767790 133686131514496 metric_logger.py:185] completed step: 8, seconds: 0.428, TFLOP/s/device: 127.005, Tokens/s/device: 19143.586, total_weights: 65536, loss: 8.251 I0416 12:51:55.190747 133686131514496 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0416 12:51:55.191635 133686131514496 checkpointing.py:776] Waited 0.0009131431579589844 seconds for step 9 to finish before starting checkpointing. I0416 12:51:55.191889 133686131514496 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 12:51:55.192061 133686131514496 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 12:51:55.192343 133686131514496 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9. I0416 12:51:55.361802 133578648192576 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9 I0416 12:51:55.369265 133686131514496 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 12:51:55.369356 133686131514496 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 I0416 12:51:56.066377 133578463643200 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9/items I0416 12:52:02.380641 133686131514496 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.011793s I0416 12:52:02.385154 133686131514496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.737 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0416 12:52:02.385221 133686131514496 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.103378s (batch_requests_ready=0.083937s, total_serialization_initiated=7.015049s, others=0.004392s) I0416 12:52:02.385334 133686131514496 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.103949s (all_items=0.000027s, per_item={'items': '0.00002718'}, temp_paths=7.103922) I0416 12:52:02.386816 133578648192576 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 12:52:02.386931 133686131514496 async_checkpointer.py:561] Finished blocking save. Time taken: 7.194829s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9. I0416 12:52:02.387180 133686131514496 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 12:52:02.387351 133578570597952 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 12:52:02.387460 133686131514496 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776343915.1918523, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': 8.564091920852661, 'checkpointer_blocking_start_time': 1776343915.1920834, 'checkpointer_blocking_duration_secs': 7.194972991943359, 'get_old_steps_start_time': 1776343922.3870852, 'get_old_steps_duration_secs': 4.7206878662109375e-05, 'checkpoint_manager_blocking_start_time': 1776343915.1918163, 'checkpoint_manager_blocking_duration_secs': 7.195617198944092} I0416 12:52:02.387610 133686131514496 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0416 12:52:02.387645 133686131514496 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 12:52:03.112103 133578637706816 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9/items/array_metadatas/process_0 I0416 12:52:36.220334 133578551723584 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.834419s (commit=33.381209s, array_metadata_write=0.453210s) I0416 12:52:36.221404 133578648192576 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 308.670 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0416 12:52:36.221513 133578648192576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.834532s. I0416 12:52:36.676491 133578648192576 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 12:52:37.054002 133578648192576 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 12:52:37.054194 133578648192576 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.532250s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9/items I0416 12:52:37.054829 133578648192576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9/items I0416 12:52:37.310606 133578648192576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9 I0416 12:52:37.974599 133578648192576 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9`. I0416 12:52:37.975217 133578648192576 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.783115s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_06_grad_accum/checkpoints/9 I0416 12:52:37.975293 133578648192576 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.588318s. I0416 12:52:37.975470 133578570597952 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 12:52:37.975583 133578570597952 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 12:52:37.975668 133578570597952 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 12:52:37.975722 133578570597952 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 12:52:37.975877 133686131514496 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. I0416 12:52:37.975998 133686131514496 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 12:52:37.976572 133686131514496 metric_logger.py:185] completed step: 9, seconds: 0.434, TFLOP/s/device: 125.115, Tokens/s/device: 18858.629, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 54.35 split as 93.93% learnable weight flops and 6.07% attention flops