2026-04-16 20:36:03.816545: 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 20:36:03.934111 136901187198080 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 20:37:00.567867 136901187198080 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 20:37:00.567972 136901187198080 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 20:37:00.568006 136901187198080 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 20:37:00.568031 136901187198080 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 20:37:00.568116 136901187198080 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 20:37:00.568193 136901187198080 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 20:37:00.660164 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:01.048309 136901187198080 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 20:37:01.146863 136901187198080 checkpointing.py:688] Setting up checkpoint logger... I0416 20:37:01.146960 136901187198080 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 20:37:01.147004 136901187198080 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 20:37:01.147207 136901187198080 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 0x7c8233bf8e60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 20:37:03.573425 136901187198080 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 20:37:03.573807 136901187198080 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 0x7c7bb6d5b860>}, handler_registry=None I0416 20:37:03.574330 136901187198080 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c7bb6d5b860>` 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 20:37:03.574377 136901187198080 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7bb6d3e720>` 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 20:37:03.574412 136901187198080 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 0x7c7bb6d5b860>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c7bb6d5b860>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7bb6d3e720>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7bb6d3e720>}). I0416 20:37:03.575245 136901187198080 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 20:37:03.575318 136901187198080 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7c7bb7d79760> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 20:37:03.702292 136901187198080 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints I0416 20:37:03.702559 136901187198080 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=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c7bb6d3c260> I0416 20:37:03.702656 136901187198080 checkpointing.py:302] Checkpoint manager created! W0416 20:37:03.721359 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:03.929949 136901187198080 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 20:37:03.930069 136901187198080 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 20:37:04.040698 4115359 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 20:37:04.128190 136901187198080 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 20:37:04.128291 136901187198080 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:37:04.128332 136901187198080 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:37:04.128366 136901187198080 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 20:37:04.128395 136901187198080 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:37:04.128420 136901187198080 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:37:04.128446 136901187198080 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:37:04.128471 136901187198080 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 20:37:04.128496 136901187198080 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:37:04.128520 136901187198080 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:37:04.128543 136901187198080 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 20:37:04.128566 136901187198080 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 20:37:04.238398 136901187198080 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:37:04.238476 136901187198080 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:37:04.243842 136901187198080 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:37:04.243889 136901187198080 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:37:04.258357 136901187198080 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:37:04.258405 136901187198080 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:37:04.272218 136901187198080 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:37:04.272263 136901187198080 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:37:04.293610 136901187198080 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 20:37:04.293673 136901187198080 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:37:04.307700 136901187198080 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 20:37:04.307750 136901187198080 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:37:04.321764 136901187198080 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 20:37:04.321812 136901187198080 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:37:04.348726 136901187198080 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 20:37:04.348785 136901187198080 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:37:04.367302 136901187198080 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 20:37:04.367362 136901187198080 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). W0416 20:37:05.095260 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:05.216303 136901187198080 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. I0416 20:37:05.216836 136901187198080 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 20:37:05.218194 136901187198080 metric_logger.py:289] number parameters: 1.104 billion W0416 20:37:05.840004 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:05.937518 136901187198080 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 20:37:06.053747 136901187198080 checkpointing.py:798] Waited 0.11622118949890137 seconds for step 0 to finish before starting checkpointing. I0416 20:37:06.054195 136901187198080 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:37:06.054346 136901187198080 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 20:37:06.054693 136901187198080 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0. I0416 20:37:06.146507 136901187198080 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 20:37:06.244017 136901187198080 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:37:06.244168 136793729009216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0 I0416 20:37:06.244561 136901187198080 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 20:37:06.886340 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.898125 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.908134 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.914759 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.934742 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:06.943464 136793718523456 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/items W0416 20:37:06.943858 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.950691 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:06.955217 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:07.095780 136793678677568 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776371826842351608, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776371826842351608, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:37:07.509841 4117857 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 20:37:07.509893 4117857 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 20:37:07.873804 136901187198080 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.631098s I0416 20:37:07.874384 136901187198080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.149 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0416 20:37:07.874439 136901187198080 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.726214s (batch_requests_ready=0.086849s, total_serialization_initiated=1.638895s, others=0.000469s) I0416 20:37:07.874506 136901187198080 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.727046s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=1.727026) I0416 20:37:07.881301 136793739494976 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:37:07.881490 136901187198080 async_checkpointer.py:561] Finished blocking save. Time taken: 1.827105s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0. I0416 20:37:07.881644 136901187198080 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:37:07.881856 136793615763008 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:37:07.881966 136901187198080 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776371826.0541837, 'wait_for_prev_duration_secs': 3.9577484130859375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776371826.0543656, 'checkpointer_blocking_duration_secs': 1.8272051811218262, 'get_old_steps_start_time': 1776371827.881589, 'get_old_steps_duration_secs': 2.0503997802734375e-05, 'checkpoint_manager_blocking_start_time': 1776371826.0541, 'checkpoint_manager_blocking_duration_secs': 1.8278343677520752} I0416 20:37:07.882090 136901187198080 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 20:37:07.882140 136901187198080 max_utils.py:750] Memstats: After params initialized: I0416 20:37:07.882185 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 20:37:07.882210 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 20:37:07.882231 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 20:37:07.882250 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 20:37:07.882267 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 20:37:07.882283 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 20:37:07.882299 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 20:37:07.882315 136901187198080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 20:37:07.889149 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:07.894705 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:37:07.899287 4115359 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:37:08.098405 136901187198080 metric_logger.py:185] completed step: 0, seconds: 0.718, TFLOP/s/device: 18.922, Tokens/s/device: 2852.169, total_weights: 16384, loss: 10.881 I0416 20:37:08.099146 136901187198080 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/tensorboard/' I0416 20:37:08.212458 136901187198080 metric_logger.py:185] completed step: 1, seconds: 2.159, TFLOP/s/device: 6.295, Tokens/s/device: 948.801, total_weights: 16384, loss: 10.881 I0416 20:37:08.331234 136901187198080 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 1334.295, Tokens/s/device: 201119.513, total_weights: 16384, loss: 10.282 I0416 20:37:08.449455 136901187198080 metric_logger.py:185] completed step: 3, seconds: 0.112, TFLOP/s/device: 121.613, Tokens/s/device: 18330.887, total_weights: 16384, loss: 9.282 I0416 20:37:08.455017 136901187198080 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0416 20:37:08.654790 136793697551936 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_0 I0416 20:37:08.686298 136901187198080 checkpointing.py:798] Waited 0.23125410079956055 seconds for step 5 to finish before starting checkpointing. I0416 20:37:08.686680 136901187198080 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:37:38.710290 136793626248768 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.829956s (commit=30.372816s, array_metadata_write=0.457139s) I0416 20:37:38.711444 136793739494976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 388.070 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 32 seconds) (per-host) I0416 20:37:38.711554 136793739494976 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.830157s. I0416 20:37:38.957017 136793739494976 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776371826842351608, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:37:39.157457 136793739494976 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 20:37:39.351212 136793678677568 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776371826842351608, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:37:39.601144 136793739494976 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:37:39.601338 136793739494976 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.603716s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/items I0416 20:37:39.601983 136793739494976 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0/items I0416 20:37:39.880036 136793739494976 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0 I0416 20:37:40.573759 136793739494976 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0`. I0416 20:37:40.574386 136793739494976 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 34.520009s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/0 I0416 20:37:40.574464 136793739494976 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.693068s. I0416 20:37:40.574649 136793615763008 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:37:40.574754 136793615763008 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:37:40.574821 136793615763008 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:37:40.574873 136793615763008 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 20:37:40.575021 136901187198080 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. W0416 20:37:40.575153 136901187198080 checkpoint_manager.py:1441] Waiting for previous save to complete took 31.888473 seconds. If this number is high, consider checkpointing less frequently. I0416 20:37:40.575899 136901187198080 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0416 20:37:40.576195 136901187198080 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5. I0416 20:37:40.758650 136793615763008 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5 I0416 20:37:40.775394 136901187198080 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:37:40.775484 136901187198080 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 20:37:41.477748 136793594791488 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5/items I0416 20:37:46.983336 136901187198080 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.208952s I0416 20:37:46.991035 136901187198080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.950 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 20:37:46.991131 136901187198080 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.330033s (batch_requests_ready=0.105871s, total_serialization_initiated=6.216612s, others=0.007550s) I0416 20:37:46.991240 136901187198080 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.330580s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=6.330562) I0416 20:37:46.993110 136793615763008 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:37:46.993235 136901187198080 async_checkpointer.py:561] Finished blocking save. Time taken: 6.417285s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5. I0416 20:37:46.993560 136901187198080 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:37:46.993785 136793739494976 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:37:46.993931 136901187198080 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776371828.686657, 'wait_for_prev_duration_secs': 31.88847327232361, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776371860.575924, 'checkpointer_blocking_duration_secs': 6.417460203170776, 'get_old_steps_start_time': 1776371866.9934177, 'get_old_steps_duration_secs': 6.961822509765625e-05, 'checkpoint_manager_blocking_start_time': 1776371828.6865969, 'checkpoint_manager_blocking_duration_secs': 38.307297229766846} I0416 20:37:46.994161 136901187198080 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0416 20:37:46.994808 136901187198080 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.717, Tokens/s/device: 17140.657, total_weights: 16384, loss: 8.323 I0416 20:37:47.007074 136901187198080 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.571, Tokens/s/device: 17420.151, total_weights: 16384, loss: 7.360 I0416 20:37:47.119544 136901187198080 metric_logger.py:185] completed step: 6, seconds: 38.548, TFLOP/s/device: 0.352, Tokens/s/device: 53.129, total_weights: 16384, loss: 6.449 I0416 20:37:47.237570 136901187198080 metric_logger.py:185] completed step: 7, seconds: 0.009, TFLOP/s/device: 1482.017, Tokens/s/device: 223385.689, total_weights: 16384, loss: 5.610 I0416 20:37:47.355882 136901187198080 metric_logger.py:185] completed step: 8, seconds: 0.115, TFLOP/s/device: 117.673, Tokens/s/device: 17736.977, total_weights: 16384, loss: 4.852 I0416 20:37:47.474730 136901187198080 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 20:37:47.476506 136901187198080 checkpointing.py:798] Waited 0.001832723617553711 seconds for step 9 to finish before starting checkpointing. I0416 20:37:47.477682 136901187198080 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:37:47.742583 136793697551936 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_0 I0416 20:38:26.544697 136793594791488 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.552765s (commit=39.077062s, array_metadata_write=0.475703s) I0416 20:38:26.546278 136793615763008 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 275.401 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0416 20:38:26.546351 136793615763008 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.553058s. I0416 20:38:27.004596 136793615763008 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 20:38:27.530460 136793615763008 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:38:27.531453 136793615763008 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.696557s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5/items I0416 20:38:27.532387 136793615763008 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5/items I0416 20:38:27.785385 136793615763008 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5 I0416 20:38:28.486017 136793615763008 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5`. I0416 20:38:28.486959 136793615763008 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.911021s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/5 I0416 20:38:28.487102 136793615763008 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.493810s. I0416 20:38:28.487305 136793739494976 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:38:28.487453 136793739494976 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:38:28.487574 136793739494976 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:38:28.487669 136793739494976 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0416 20:38:28.488415 136901187198080 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0416 20:38:28.489273 136901187198080 checkpoint_manager.py:1441] Waiting for previous save to complete took 41.011593 seconds. If this number is high, consider checkpointing less frequently. I0416 20:38:28.489960 136901187198080 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 20:38:28.490414 136901187198080 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9. I0416 20:38:28.686923 136901187198080 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:38:28.687162 136793739494976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9 I0416 20:38:28.687294 136901187198080 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 20:38:29.420992 136793636734528 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9/items I0416 20:38:38.943172 136901187198080 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.258029s I0416 20:38:38.943955 136901187198080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.191 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0416 20:38:38.944014 136901187198080 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.361249s (batch_requests_ready=0.093562s, total_serialization_initiated=10.267113s, others=0.000573s) I0416 20:38:38.944113 136901187198080 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.361980s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=10.361962) I0416 20:38:38.954176 136793739494976 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:38:38.954349 136901187198080 async_checkpointer.py:561] Finished blocking save. Time taken: 10.464332s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9. I0416 20:38:38.954587 136901187198080 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:38:38.954797 136793615763008 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:38:38.954906 136901187198080 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776371867.4776487, 'wait_for_prev_duration_secs': 41.011592626571655, 'time_between_consecutive_saves_sec': 6.902760982513428, 'checkpointer_blocking_start_time': 1776371908.4899876, 'checkpointer_blocking_duration_secs': 10.46451997756958, 'get_old_steps_start_time': 1776371918.9545248, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776371867.4776053, 'checkpoint_manager_blocking_duration_secs': 51.4772675037384} I0416 20:38:38.955028 136901187198080 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 20:38:38.955082 136901187198080 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:38:39.765170 136793697551936 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_0 I0416 20:39:07.556753 136793594791488 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 28.603498s (commit=28.123551s, array_metadata_write=0.479948s) I0416 20:39:07.559205 136793739494976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 324.217 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0416 20:39:07.559354 136793739494976 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.604945s. I0416 20:39:07.980784 136793739494976 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 20:39:08.402704 136793739494976 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:39:08.403708 136793739494976 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570870s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9/items I0416 20:39:08.404646 136793739494976 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9/items I0416 20:39:08.657761 136793739494976 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9 I0416 20:39:09.380993 136793739494976 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9`. I0416 20:39:09.381891 136793739494976 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.891890s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_true/checkpoints/9 I0416 20:39:09.381967 136793739494976 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 30.427560s. I0416 20:39:09.382144 136793615763008 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:39:09.382205 136793615763008 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:39:09.382307 136793615763008 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:39:09.382357 136793615763008 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 20:39:09.383030 136901187198080 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 20:39:09.383261 136901187198080 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:39:09.384446 136901187198080 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 116.135, Tokens/s/device: 17505.171, total_weights: 16384, loss: 4.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops