2026-04-15 23:43:56.470833: 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 23:43:56.586633 125870343748736 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 23:44:51.912612 125870343748736 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 23:44:51.912728 125870343748736 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 23:44:51.912763 125870343748736 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 23:44:51.912794 125870343748736 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 23:44:51.912819 125870343748736 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 23:44:51.912918 125870343748736 train.py:715] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 23:44:52.005997 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:52.410352 125870343748736 maxtext_utils.py:1520] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 23:44:52.410708 125870343748736 checkpointing.py:677] Setting up checkpoint logger... I0415 23:44:52.410757 125870343748736 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 23:44:52.410823 125870343748736 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 23:44:52.411531 125870343748736 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 0x7279e2300b60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 23:44:54.810708 125870343748736 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 23:44:54.810921 125870343748736 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 0x7273655abb60>}, handler_registry=None I0415 23:44:54.811256 125870343748736 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7273655abb60>` 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 23:44:54.811299 125870343748736 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72736453e480>` 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 23:44:54.811327 125870343748736 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 0x7273655abb60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7273655abb60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72736453e480>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72736453e480>}). I0415 23:44:54.811659 125870343748736 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 23:44:54.811717 125870343748736 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7273645394e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 23:44:54.940961 125870343748736 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints I0415 23:44:54.941225 125870343748736 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_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x72736453e120> I0415 23:44:54.941313 125870343748736 checkpointing.py:301] Checkpoint manager created! W0415 23:44:54.960569 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:55.140005 125870343748736 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 23:44:55.140186 125870343748736 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:44:55.232617 125870343748736 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:44:55.232692 125870343748736 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:44:55.247081 125870343748736 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:44:55.247128 125870343748736 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:44:55.267977 125870343748736 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 23:44:55.268031 125870343748736 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:44:55.282655 125870343748736 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 23:44:55.282705 125870343748736 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:44:55.297089 125870343748736 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 23:44:55.297136 125870343748736 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:44:55.318965 125870343748736 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 23:44:55.319022 125870343748736 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:44:55.337618 125870343748736 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 23:44:55.337668 125870343748736 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 23:44:55.522948 125870343748736 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 23:44:55.523060 125870343748736 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0415 23:44:55.573189 3389848 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 23:44:55.955929 125870343748736 maxtext_utils.py:1623] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 23:44:55.956034 125870343748736 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:44:55.956125 125870343748736 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:44:55.956171 125870343748736 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 23:44:55.956212 125870343748736 maxtext_utils.py:1623] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:44:55.956240 125870343748736 maxtext_utils.py:1623] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:44:55.956283 125870343748736 maxtext_utils.py:1623] 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 23:44:55.956322 125870343748736 maxtext_utils.py:1623] 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 23:44:55.956351 125870343748736 maxtext_utils.py:1623] 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 23:44:55.956377 125870343748736 maxtext_utils.py:1623] 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 23:44:55.956410 125870343748736 maxtext_utils.py:1623] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed', 'vocab') Physical: ('fsdp', None) I0415 23:44:55.956443 125870343748736 maxtext_utils.py:1623] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed') Physical: (None, 'fsdp') I0415 23:44:56.380748 125870343748736 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:44:56.380830 125870343748736 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 23:44:56.394451 125870343748736 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:44:56.394502 125870343748736 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 23:44:57.059080 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:57.186729 125870343748736 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0415 23:44:57.187222 125870343748736 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 23:44:57.187363 125870343748736 metric_logger.py:289] number parameters: 1.104 billion W0415 23:44:58.039408 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:58.135886 125870343748736 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0415 23:44:58.252933 125870343748736 checkpointing.py:776] Waited 0.1170358657836914 seconds for step 0 to finish before starting checkpointing. I0415 23:44:58.253389 125870343748736 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:44:58.253540 125870343748736 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0415 23:44:58.253920 125870343748736 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0. I0415 23:44:58.337038 125870343748736 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 23:44:58.409771 125762854192704 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0 I0415 23:44:58.442227 125870343748736 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:44:58.442317 125870343748736 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 W0415 23:44:58.656307 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.671126 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.684723 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.703530 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.711421 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.716936 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.721860 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:58.726594 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:59.052378 125762843706944 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/items I0415 23:44:59.141651 125870343748736 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.699845s I0415 23:44:59.142076 125870343748736 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 15.354 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 803 milliseconds) (per-host) I0415 23:44:59.142134 125870343748736 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.803862s (batch_requests_ready=0.100185s, total_serialization_initiated=0.703336s, others=0.000340s) I0415 23:44:59.142194 125870343748736 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.804699s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.804679) I0415 23:44:59.142939 125762791278144 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:44:59.143030 125870343748736 async_checkpointer.py:561] Finished blocking save. Time taken: 0.889451s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0. I0415 23:44:59.143227 125870343748736 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:44:59.143432 125870343748736 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776296698.2533755, 'wait_for_prev_duration_secs': 4.172325134277344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776296698.2535608, 'checkpointer_blocking_duration_secs': 0.889594316482544, 'get_old_steps_start_time': 1776296699.1431723, 'get_old_steps_duration_secs': 2.0503997802734375e-05, 'checkpoint_manager_blocking_start_time': 1776296698.2532928, 'checkpoint_manager_blocking_duration_secs': 0.8901174068450928} I0415 23:44:59.143499 125870343748736 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0415 23:44:59.143535 125870343748736 max_utils.py:750] Memstats: After params initialized: I0415 23:44:59.143605 125870343748736 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0415 23:44:59.143630 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 23:44:59.143650 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 23:44:59.143667 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 23:44:59.143685 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 23:44:59.143702 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 23:44:59.143718 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 23:44:59.143739 125870343748736 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 23:44:59.149349 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:59.155121 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:44:59.160144 3389848 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:44:59.242901 125762812249664 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776296698969506234, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776296698969506234, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:44:59.243123 125762864678464 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:44:59.498947 125870343748736 metric_logger.py:185] completed step: 0, seconds: 0.948, TFLOP/s/device: 14.332, Tokens/s/device: 2160.212, total_weights: 16384, loss: 10.874 I0415 23:44:59.499955 125870343748736 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/tensorboard/' I0415 23:44:59.529410 3392128 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 23:44:59.529470 3392128 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 23:44:59.614959 125870343748736 metric_logger.py:185] completed step: 1, seconds: 1.361, TFLOP/s/device: 9.982, Tokens/s/device: 1504.618, total_weights: 16384, loss: 10.874 I0415 23:44:59.733771 125870343748736 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 2039.803, Tokens/s/device: 307461.342, total_weights: 16384, loss: 9.775 I0415 23:44:59.852710 125870343748736 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.173, Tokens/s/device: 17510.859, total_weights: 16384, loss: 8.838 I0415 23:45:00.264013 125762833221184 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0415 23:45:19.432490 125870343748736 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 114.661, Tokens/s/device: 17282.992, total_weights: 16384, loss: 7.965 I0415 23:45:19.440302 125870343748736 metric_logger.py:185] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.093, Tokens/s/device: 17197.367, total_weights: 16384, loss: 7.233 I0415 23:45:19.557549 125870343748736 metric_logger.py:185] completed step: 6, seconds: 19.581, TFLOP/s/device: 0.694, Tokens/s/device: 104.591, total_weights: 16384, loss: 6.676 I0415 23:45:19.676121 125870343748736 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 2867.088, Tokens/s/device: 432158.683, total_weights: 16384, loss: 6.293 I0415 23:45:19.794342 125870343748736 metric_logger.py:185] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.139, Tokens/s/device: 17355.050, total_weights: 16384, loss: 6.055 I0415 23:45:19.912374 125870343748736 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0415 23:45:19.913126 125870343748736 checkpointing.py:776] Waited 0.0007712841033935547 seconds for step 9 to finish before starting checkpointing. I0415 23:45:19.913392 125870343748736 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:45:34.704913 125762801763904 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.172695s (commit=34.751570s, array_metadata_write=0.421126s) I0415 23:45:34.706093 125762791278144 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 347.472 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0415 23:45:34.706204 125762791278144 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.563115s. I0415 23:45:34.940174 125762791278144 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776296698969506234, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:45:35.118649 125762791278144 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 23:45:35.325649 125762812249664 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776296698969506234, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:45:35.553371 125762791278144 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:45:35.554023 125762791278144 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.575290s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/items I0415 23:45:35.554759 125762791278144 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0/items I0415 23:45:35.792165 125762791278144 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0 I0415 23:45:36.459904 125762791278144 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0`. I0415 23:45:36.460531 125762791278144 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.206959s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/0 I0415 23:45:36.460644 125762791278144 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.317556s. I0415 23:45:36.460835 125762864678464 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:45:36.460947 125762864678464 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:45:36.461005 125762864678464 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:45:36.461062 125762864678464 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0415 23:45:36.461167 125870343748736 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 23:45:36.461249 125870343748736 checkpoint_manager.py:1441] Waiting for previous save to complete took 16.547878 seconds. If this number is high, consider checkpointing less frequently. I0415 23:45:36.462037 125870343748736 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0415 23:45:36.462288 125870343748736 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9. I0415 23:45:36.615762 125762864678464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9 I0415 23:45:36.623818 125870343748736 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:45:36.623920 125870343748736 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 23:45:37.304555 125762648671808 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9/items I0415 23:45:46.536882 125870343748736 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.913444s I0415 23:45:46.541296 125870343748736 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.233 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0415 23:45:46.541363 125870343748736 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.008910s (batch_requests_ready=0.087969s, total_serialization_initiated=9.916627s, others=0.004315s) I0415 23:45:46.541437 125870343748736 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.009422s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=10.009406) I0415 23:45:46.542827 125762864678464 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:45:46.542908 125870343748736 async_checkpointer.py:561] Finished blocking save. Time taken: 10.080811s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9. I0415 23:45:46.543108 125870343748736 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:45:46.543256 125762791278144 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:45:46.543356 125870343748736 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776296719.9133475, 'wait_for_prev_duration_secs': 16.5478777885437, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776296736.4620771, 'checkpointer_blocking_duration_secs': 10.080939531326294, 'get_old_steps_start_time': 1776296746.5430307, 'get_old_steps_duration_secs': 3.886222839355469e-05, 'checkpoint_manager_blocking_start_time': 1776296719.9133055, 'checkpoint_manager_blocking_duration_secs': 26.630025386810303} I0415 23:45:46.543473 125870343748736 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0415 23:45:46.543506 125870343748736 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:45:47.314828 125762833221184 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0415 23:46:20.851127 125762648671808 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.309099s (commit=33.854044s, array_metadata_write=0.455055s) I0415 23:46:20.852085 125762864678464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 285.129 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0415 23:46:20.852146 125762864678464 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.309189s. I0415 23:46:21.285603 125762864678464 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 23:46:21.713535 125762864678464 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:46:21.713717 125762864678464 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.564123s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9/items I0415 23:46:21.714378 125762864678464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9/items I0415 23:46:21.954532 125762864678464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9 I0415 23:46:22.630349 125762864678464 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9`. I0415 23:46:22.630982 125762864678464 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.168887s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_11_optimizer_offload_false/checkpoints/9 I0415 23:46:22.631074 125762864678464 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.088123s. I0415 23:46:22.631236 125762791278144 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:46:22.631340 125762791278144 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:46:22.631417 125762791278144 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:46:22.631469 125762791278144 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0415 23:46:22.631722 125870343748736 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 23:46:22.632169 125870343748736 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:46:22.632708 125870343748736 metric_logger.py:185] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.534, Tokens/s/device: 17263.761, 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