2026-04-16 20:57:23.899214: 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 20:57:24.016704 123946965953664 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 20:58:21.921158 123946965953664 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 20:58:21.921276 123946965953664 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 20:58:21.921312 123946965953664 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 20:58:21.921336 123946965953664 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 20:58:21.921357 123946965953664 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 20:58:21.921440 123946965953664 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 20:58:22.014611 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:22.416805 123946965953664 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 20:58:22.862186 123946965953664 checkpointing.py:688] Setting up checkpoint logger... I0416 20:58:22.862283 123946965953664 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0416 20:58:22.862320 123946965953664 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 20:58:22.862488 123946965953664 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x70ba0ff01070>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 20:58:25.204210 123946965953664 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 20:58:25.204378 123946965953664 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 0x70b511b415e0>}, handler_registry=None I0416 20:58:25.204588 123946965953664 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70b511b415e0>` 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 20:58:25.204627 123946965953664 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70b511c8a510>` 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 20:58:25.204655 123946965953664 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 0x70b511b415e0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70b511b415e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70b511c8a510>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70b511c8a510>}). I0416 20:58:25.204959 123946965953664 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 20:58:25.205008 123946965953664 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x70b3955af240> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 20:58:25.318397 123946965953664 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_16_zarr3_false/checkpoints I0416 20:58:25.318606 123946965953664 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_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x70b511c8ad80> I0416 20:58:25.318683 123946965953664 checkpointing.py:302] Checkpoint manager created! W0416 20:58:25.334317 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:25.541795 123946965953664 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 20:58:25.541899 123946965953664 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 20:58:25.651876 4138459 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 20:58:25.738271 123946965953664 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 20:58:25.738368 123946965953664 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:58:25.738409 123946965953664 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:58:25.738441 123946965953664 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 20:58:25.738470 123946965953664 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:58:25.738494 123946965953664 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:58:25.738522 123946965953664 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:58:25.738546 123946965953664 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 20:58:25.738571 123946965953664 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:58:25.738594 123946965953664 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:58:25.738616 123946965953664 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 20:58:25.738639 123946965953664 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 20:58:25.834879 123946965953664 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:58:25.834961 123946965953664 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:58:25.840086 123946965953664 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:58:25.840131 123946965953664 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:58:25.854208 123946965953664 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:58:25.854255 123946965953664 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:58:25.867768 123946965953664 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:58:25.867813 123946965953664 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:58:25.888655 123946965953664 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 20:58:25.888709 123946965953664 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:58:25.902396 123946965953664 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 20:58:25.902445 123946965953664 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:58:25.916032 123946965953664 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 20:58:25.916096 123946965953664 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:58:25.943812 123946965953664 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 20:58:25.943870 123946965953664 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:58:25.960950 123946965953664 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 20:58:25.961000 123946965953664 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). W0416 20:58:26.305374 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:26.422555 123946965953664 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 20:58:26.423114 123946965953664 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 20:58:26.424523 123946965953664 metric_logger.py:289] number parameters: 1.104 billion W0416 20:58:27.021309 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:27.141904 123946965953664 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 20:58:27.258233 123946965953664 checkpointing.py:798] Waited 0.11632561683654785 seconds for step 0 to finish before starting checkpointing. I0416 20:58:27.258654 123946965953664 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:58:27.258802 123946965953664 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 20:58:27.259177 123946965953664 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_16_zarr3_false/checkpoints/0. I0416 20:58:27.344688 123946965953664 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 20:58:27.445118 123839518344768 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_16_zarr3_false/checkpoints/0 I0416 20:58:27.445592 123946965953664 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:58:27.445810 123946965953664 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 20:58:28.092826 123839507859008 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_16_zarr3_false/checkpoints/0/items W0416 20:58:28.093404 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.102668 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.114150 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.125766 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.135659 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.142854 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.148127 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.152745 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:28.238093 123839476401728 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776373107997846838, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776373107997846838, "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_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:58:28.592356 123946965953664 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.148284s I0416 20:58:28.592791 123946965953664 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 9.897 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0416 20:58:28.592841 123946965953664 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.246950s (batch_requests_ready=0.089679s, total_serialization_initiated=1.156928s, others=0.000344s) I0416 20:58:28.592902 123946965953664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.247479s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=1.247460) I0416 20:58:28.593716 123839455430208 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:58:28.593814 123946965953664 async_checkpointer.py:561] Finished blocking save. Time taken: 1.334975s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/0. I0416 20:58:28.594001 123946965953664 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:58:28.594270 123839528830528 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:58:28.594372 123946965953664 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776373107.2586422, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776373107.2588217, 'checkpointer_blocking_duration_secs': 1.335108995437622, 'get_old_steps_start_time': 1776373108.593947, 'get_old_steps_duration_secs': 2.0265579223632812e-05, 'checkpoint_manager_blocking_start_time': 1776373107.258562, 'checkpoint_manager_blocking_duration_secs': 1.335784912109375} I0416 20:58:28.594481 123946965953664 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 20:58:28.594532 123946965953664 max_utils.py:750] Memstats: After params initialized: I0416 20:58:28.594629 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 20:58:28.594655 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 20:58:28.594675 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 20:58:28.594693 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 20:58:28.594711 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 20:58:28.594728 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 20:58:28.594745 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 20:58:28.594763 123946965953664 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 20:58:28.600351 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:58:28.606231 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:28.614549 4141090 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 20:58:28.614613 4141090 google_auth_provider.cc:156] Using OAuth2 AuthProvider W0416 20:58:28.618704 4138459 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:58:28.816586 123946965953664 metric_logger.py:185] completed step: 0, seconds: 0.716, TFLOP/s/device: 18.973, Tokens/s/device: 2859.868, total_weights: 16384, loss: 10.881 I0416 20:58:28.817479 123946965953664 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_16_zarr3_false/tensorboard/' I0416 20:58:28.930595 123946965953664 metric_logger.py:185] completed step: 1, seconds: 1.672, TFLOP/s/device: 8.125, Tokens/s/device: 1224.696, total_weights: 16384, loss: 10.881 I0416 20:58:29.049231 123946965953664 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 1309.477, Tokens/s/device: 197378.566, total_weights: 16384, loss: 10.282 I0416 20:58:29.167743 123946965953664 metric_logger.py:185] completed step: 3, seconds: 0.112, TFLOP/s/device: 121.672, Tokens/s/device: 18339.751, total_weights: 16384, loss: 9.282 I0416 20:58:29.173395 123946965953664 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0416 20:58:29.347889 123839497373248 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_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0416 20:58:29.404449 123946965953664 checkpointing.py:798] Waited 0.23103070259094238 seconds for step 5 to finish before starting checkpointing. I0416 20:58:29.405385 123946965953664 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:59:09.585433 123839518344768 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.977858s (commit=40.568652s, array_metadata_write=0.409207s) I0416 20:59:09.586610 123839455430208 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 299.162 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 42 seconds) (per-host) I0416 20:59:09.586725 123839455430208 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.992865s. I0416 20:59:09.839298 123839455430208 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776373107997846838, '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_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:59:10.017885 123839455430208 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 20:59:10.236418 123839476401728 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776373107997846838, '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_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:59:10.483856 123839455430208 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.602031s. use_zarr3=False, 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_16_zarr3_false/checkpoints/0/items I0416 20:59:10.484570 123839455430208 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/0/items I0416 20:59:10.754107 123839455430208 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/0 I0416 20:59:11.427749 123839455430208 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_16_zarr3_false/checkpoints/0`. I0416 20:59:11.428372 123839455430208 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 44.169539s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/0 I0416 20:59:11.428447 123839455430208 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 42.834589s. I0416 20:59:11.428648 123839528830528 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:59:11.428759 123839528830528 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:59:11.428831 123839528830528 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:59:11.428879 123839528830528 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 20:59:11.429060 123946965953664 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 20:59:11.429183 123946965953664 checkpoint_manager.py:1441] Waiting for previous save to complete took 42.023797 seconds. If this number is high, consider checkpointing less frequently. I0416 20:59:11.429932 123946965953664 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0416 20:59:11.430222 123946965953664 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_16_zarr3_false/checkpoints/5. I0416 20:59:11.603993 123839528830528 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_16_zarr3_false/checkpoints/5 I0416 20:59:11.635931 123946965953664 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:59:11.636024 123946965953664 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 20:59:12.290930 123839465915968 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_16_zarr3_false/checkpoints/5/items I0416 20:59:24.024502 123946965953664 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.389564s I0416 20:59:24.030323 123946965953664 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1010.424 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0416 20:59:24.030385 123946965953664 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.506517s (batch_requests_ready=0.103916s, total_serialization_initiated=12.396887s, others=0.005714s) I0416 20:59:24.030467 123946965953664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.507005s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=12.506989) I0416 20:59:24.031849 123839413487168 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:59:24.032022 123946965953664 async_checkpointer.py:561] Finished blocking save. Time taken: 12.602047s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/5. I0416 20:59:24.032230 123946965953664 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:59:24.032452 123839392515648 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:59:24.032580 123946965953664 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776373109.4053626, 'wait_for_prev_duration_secs': 42.023796796798706, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776373151.429957, 'checkpointer_blocking_duration_secs': 12.60218620300293, 'get_old_steps_start_time': 1776373164.0321627, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776373109.4053004, 'checkpoint_manager_blocking_duration_secs': 54.627254009246826} I0416 20:59:24.032691 123946965953664 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0416 20:59:24.033251 123946965953664 metric_logger.py:185] completed step: 4, seconds: 0.120, TFLOP/s/device: 113.217, Tokens/s/device: 17065.245, total_weights: 16384, loss: 8.323 I0416 20:59:24.043460 123946965953664 metric_logger.py:185] completed step: 5, seconds: 0.117, TFLOP/s/device: 115.970, Tokens/s/device: 17480.220, total_weights: 16384, loss: 7.360 I0416 20:59:24.156693 123946965953664 metric_logger.py:185] completed step: 6, seconds: 54.867, TFLOP/s/device: 0.248, Tokens/s/device: 37.327, total_weights: 16384, loss: 6.449 I0416 20:59:24.274874 123946965953664 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1714.464, Tokens/s/device: 258422.713, total_weights: 16384, loss: 5.610 I0416 20:59:24.393154 123946965953664 metric_logger.py:185] completed step: 8, seconds: 0.115, TFLOP/s/device: 118.262, Tokens/s/device: 17825.746, total_weights: 16384, loss: 4.852 I0416 20:59:24.512399 123946965953664 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 20:59:24.513610 123946965953664 checkpointing.py:798] Waited 0.0012285709381103516 seconds for step 9 to finish before starting checkpointing. I0416 20:59:24.514428 123946965953664 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:59:24.808850 123839518344768 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_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0416 21:00:02.495015 123839342184000 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.463895s (commit=38.008984s, array_metadata_write=0.454911s) I0416 21:00:02.496243 123839413487168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 247.915 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0416 21:00:02.496349 123839413487168 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.464362s. I0416 21:00:02.914529 123839413487168 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 21:00:03.563581 123839413487168 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.788108s. use_zarr3=False, 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_16_zarr3_false/checkpoints/5/items I0416 21:00:03.564311 123839413487168 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/5/items I0416 21:00:03.799423 123839413487168 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/5 I0416 21:00:04.636098 123839413487168 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_16_zarr3_false/checkpoints/5`. I0416 21:00:04.636717 123839413487168 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 53.206748s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/5 I0416 21:00:04.636789 123839413487168 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.604806s. I0416 21:00:04.636962 123839392515648 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 21:00:04.637073 123839392515648 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 21:00:04.637147 123839392515648 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0416 21:00:04.637194 123839392515648 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0416 21:00:04.637337 123946965953664 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 21:00:04.637454 123946965953664 checkpoint_manager.py:1441] Waiting for previous save to complete took 40.123031 seconds. If this number is high, consider checkpointing less frequently. I0416 21:00:04.637700 123946965953664 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 21:00:04.637995 123946965953664 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_16_zarr3_false/checkpoints/9. I0416 21:00:04.812350 123946965953664 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 21:00:04.812446 123946965953664 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 21:00:04.815061 123839392515648 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_16_zarr3_false/checkpoints/9 I0416 21:00:05.547233 123839403001408 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_16_zarr3_false/checkpoints/9/items I0416 21:00:09.593283 123946965953664 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.781958s I0416 21:00:09.598544 123946965953664 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.528 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0416 21:00:09.598607 123946965953664 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.881733s (batch_requests_ready=0.086657s, total_serialization_initiated=4.789915s, others=0.005161s) I0416 21:00:09.598682 123946965953664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.882250s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=4.882237) I0416 21:00:09.600057 123839392515648 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 21:00:09.600141 123946965953664 async_checkpointer.py:561] Finished blocking save. Time taken: 4.962398s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/9. I0416 21:00:09.600313 123946965953664 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 21:00:09.600507 123839465915968 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 21:00:09.600620 123946965953664 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776373164.514404, 'wait_for_prev_duration_secs': 40.12303137779236, 'time_between_consecutive_saves_sec': 13.085509538650513, 'checkpointer_blocking_start_time': 1776373204.6377249, 'checkpointer_blocking_duration_secs': 4.962517738342285, 'get_old_steps_start_time': 1776373209.6002574, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1776373164.5143723, 'checkpoint_manager_blocking_duration_secs': 45.08622455596924} I0416 21:00:09.600746 123946965953664 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 21:00:09.600778 123946965953664 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 21:00:10.368420 123839528830528 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_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0416 21:00:46.511365 123839342184000 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.912087s (commit=36.471717s, array_metadata_write=0.440371s) I0416 21:00:46.512467 123839392515648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 302.347 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0416 21:00:46.512522 123839392515648 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.912338s. I0416 21:00:46.921188 123839392515648 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 21:00:47.448888 123839392515648 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.666448s. use_zarr3=False, 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_16_zarr3_false/checkpoints/9/items I0416 21:00:47.449604 123839392515648 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/9/items I0416 21:00:47.694207 123839392515648 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/9 I0416 21:00:48.412483 123839392515648 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_16_zarr3_false/checkpoints/9`. I0416 21:00:48.413133 123839392515648 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.775396s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_false/checkpoints/9 I0416 21:00:48.413197 123839392515648 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.813015s. I0416 21:00:48.413350 123839465915968 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 21:00:48.413449 123839465915968 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 21:00:48.413501 123839465915968 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 21:00:48.413539 123839465915968 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 21:00:48.413672 123946965953664 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 21:00:48.413848 123946965953664 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 21:00:48.414728 123946965953664 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 115.845, Tokens/s/device: 17461.441, 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