2026-04-15 23:55:09.663346: 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 23:55:09.780870 125334245051520 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 23:55:52.639424 125334245051520 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 23:55:52.639531 125334245051520 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 23:55:52.639564 125334245051520 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 23:55:52.639589 125334245051520 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 23:55:52.639616 125334245051520 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 23:55:52.639701 125334245051520 train.py:715] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 23:55:52.733784 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:55:53.138803 125334245051520 maxtext_utils.py:1520] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 23:55:53.139152 125334245051520 checkpointing.py:677] Setting up checkpoint logger... I0415 23:55:53.139199 125334245051520 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 23:55:53.139241 125334245051520 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 23:55:53.139920 125334245051520 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 0x71fd10dfee10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 23:55:55.537339 125334245051520 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 23:55:55.537628 125334245051520 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 0x71f76452bc20>}, handler_registry=None I0415 23:55:55.537961 125334245051520 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71f76452bc20>` 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 23:55:55.538004 125334245051520 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f69593a600>` 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 23:55:55.538036 125334245051520 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 0x71f76452bc20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71f76452bc20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f69593a600>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f69593a600>}). I0415 23:55:55.538400 125334245051520 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 23:55:55.662267 125334245051520 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_14_async_ckpt_false/checkpoints I0415 23:55:55.662494 125334245051520 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=False, 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/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x71f695938e00> I0415 23:55:55.662573 125334245051520 checkpointing.py:301] Checkpoint manager created! W0415 23:55:55.680316 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:55:55.859848 125334245051520 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 23:55:55.859953 125334245051520 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:55:55.952550 125334245051520 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:55:55.952627 125334245051520 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:55:55.966708 125334245051520 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:55:55.966756 125334245051520 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:55:55.986951 125334245051520 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 23:55:55.987007 125334245051520 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:55:56.001040 125334245051520 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 23:55:56.001099 125334245051520 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:55:56.015111 125334245051520 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 23:55:56.015160 125334245051520 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:55:56.036520 125334245051520 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 23:55:56.036576 125334245051520 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:55:56.054676 125334245051520 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 23:55:56.054728 125334245051520 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 23:55:56.238220 125334245051520 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 23:55:56.238319 125334245051520 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0415 23:55:56.288700 3401420 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 23:55:56.666276 125334245051520 maxtext_utils.py:1623] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 23:55:56.666381 125334245051520 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 23:55:56.666422 125334245051520 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 23:55:56.666466 125334245051520 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 23:55:56.666509 125334245051520 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 23:55:56.666539 125334245051520 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 23:55:56.666579 125334245051520 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 23:55:56.666616 125334245051520 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 23:55:56.666644 125334245051520 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 23:55:56.666670 125334245051520 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 23:55:56.666704 125334245051520 maxtext_utils.py:1623] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed', 'vocab') Physical: ('fsdp', None) I0415 23:55:56.666737 125334245051520 maxtext_utils.py:1623] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed') Physical: (None, 'fsdp') I0415 23:55:57.084338 125334245051520 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:55:57.084418 125334245051520 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 23:55:57.097750 125334245051520 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:55:57.097801 125334245051520 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 23:55:57.751502 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:55:57.870089 125334245051520 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 23:55:57.870553 125334245051520 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 23:55:57.870697 125334245051520 metric_logger.py:289] number parameters: 1.104 billion W0415 23:55:58.717282 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:55:58.815679 125334245051520 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0415 23:55:58.932478 125334245051520 checkpointing.py:776] Waited 0.11678624153137207 seconds for step 0 to finish before starting checkpointing. I0415 23:55:58.932904 125334245051520 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:55:58.933069 125334245051520 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0415 23:55:58.933128 125334245051520 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0. I0415 23:55:59.133754 125334245051520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0 I0415 23:55:59.915654 125334245051520 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297359680267563, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776297359680267563, "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_14_async_ckpt_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:55:59.916687 125334245051520 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient W0415 23:55:59.916984 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:56:00.000004 125226662757952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0/items I0415 23:56:00.555891 125334245051520 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:56:00.555994 125334245051520 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 23:56:00.572597 3403388 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 23:56:00.572642 3403388 google_auth_provider.cc:156] Using OAuth2 AuthProvider W0415 23:56:00.781076 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.795922 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.805662 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.816915 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.824361 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.837146 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.842838 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:00.847838 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:56:01.035255 125334245051520 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.479724s I0415 23:56:01.035845 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 21.393 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 576 milliseconds) (per-host) I0415 23:56:01.040071 125334245051520 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.581140s (batch_requests_ready=0.093294s, total_serialization_initiated=0.483296s, others=0.004551s) I0415 23:56:01.040325 125334245051520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.123693s (all_items=0.000031s, per_item={'items': '0.00003099'}, temp_paths=1.123662) W0415 23:56:01.040537 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:56:01.796013 125226673243712 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_14_async_ckpt_false/checkpoints/0/items/array_metadatas/process_0 I0415 23:56:23.505888 125232503326272 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 22.465655s (commit=22.019358s, array_metadata_write=0.446296s) I0415 23:56:23.506982 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 548.281 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 23 seconds) (per-host) I0415 23:56:23.718976 125334245051520 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297359680267563, '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_14_async_ckpt_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:56:24.101272 125334245051520 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776297359680267563, '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_14_async_ckpt_false/checkpoints/0/_CHECKPOINT_METADATA I0415 23:56:24.272252 125334245051520 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:56:24.659974 125334245051520 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:56:24.660147 125334245051520 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.558487s. 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_14_async_ckpt_false/checkpoints/0/items I0415 23:56:24.660776 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0/items I0415 23:56:24.909642 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0 I0415 23:56:25.568414 125334245051520 atomicity.py:794] [process=0][thread=MainThread] 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_14_async_ckpt_false/checkpoints/0`. I0415 23:56:25.569426 125334245051520 checkpointer.py:290] Finished synchronous save in 26.64 seconds to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/0 I0415 23:56:25.569617 125334245051520 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:56:25.569662 125334245051520 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=0] CheckpointManager Save Finalize is done on all hosts. I0415 23:56:25.569687 125334245051520 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=0] Finished synchronous save. I0415 23:56:25.569724 125334245051520 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_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776297358.9328916, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776297358.933091, 'checkpointer_blocking_duration_secs': 26.636383295059204, 'get_old_steps_start_time': 1776297385.5695064, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776297358.932802, 'checkpoint_manager_blocking_duration_secs': 26.63689875602722} I0415 23:56:25.569824 125334245051520 checkpointing.py:410] Saved a checkpoint at step 0. I0415 23:56:25.569873 125334245051520 max_utils.py:750] Memstats: After params initialized: I0415 23:56:25.569927 125334245051520 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0415 23:56:25.569954 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 23:56:25.569977 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 23:56:25.569996 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 23:56:25.570013 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 23:56:25.570030 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 23:56:25.570063 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 23:56:25.570087 125334245051520 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 23:56:25.576111 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:25.581641 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:56:25.586120 3401420 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:56:25.774750 125334245051520 metric_logger.py:185] completed step: 0, seconds: 0.945, TFLOP/s/device: 14.385, Tokens/s/device: 2168.334, total_weights: 16384, loss: 10.874 I0415 23:56:25.775658 125334245051520 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_14_async_ckpt_false/tensorboard/' I0415 23:56:25.889602 125334245051520 metric_logger.py:185] completed step: 1, seconds: 26.956, TFLOP/s/device: 0.504, Tokens/s/device: 75.975, total_weights: 16384, loss: 10.874 I0415 23:56:26.008170 125334245051520 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 2017.391, Tokens/s/device: 304083.148, total_weights: 16384, loss: 10.276 I0415 23:56:26.126028 125334245051520 metric_logger.py:185] completed step: 3, seconds: 0.116, TFLOP/s/device: 117.471, Tokens/s/device: 17706.460, total_weights: 16384, loss: 9.282 I0415 23:56:26.130413 125334245051520 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0415 23:56:26.361984 125334245051520 checkpointing.py:776] Waited 0.2315373420715332 seconds for step 5 to finish before starting checkpointing. I0415 23:56:26.362386 125334245051520 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:56:26.362538 125334245051520 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0415 23:56:26.362595 125334245051520 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5. I0415 23:56:26.537604 125334245051520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5 W0415 23:56:27.366454 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:56:27.455174 125226127984192 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5/items I0415 23:56:28.001304 125334245051520 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:56:28.001409 125334245051520 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 23:56:28.158215 125334245051520 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.157273s I0415 23:56:28.158602 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 51.133 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 241 milliseconds) (per-host) I0415 23:56:28.158655 125334245051520 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.241438s (batch_requests_ready=0.080569s, total_serialization_initiated=0.160602s, others=0.000268s) I0415 23:56:28.158721 125334245051520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.792581s (all_items=0.000033s, per_item={'items': '0.00003290'}, temp_paths=0.792548) W0415 23:56:28.158754 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:56:28.962784 125225276540480 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_14_async_ckpt_false/checkpoints/5/items/array_metadatas/process_0 I0415 23:56:53.312453 125225266054720 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 25.149667s (commit=24.646396s, array_metadata_write=0.503270s) I0415 23:56:53.313604 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 497.583 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host) I0415 23:56:54.108265 125334245051520 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:56:54.552067 125334245051520 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:56:54.552239 125334245051520 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.602037s. 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_14_async_ckpt_false/checkpoints/5/items I0415 23:56:54.552896 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5/items I0415 23:56:54.792763 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5 I0415 23:56:55.458242 125334245051520 atomicity.py:794] [process=0][thread=MainThread] 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_14_async_ckpt_false/checkpoints/5`. I0415 23:56:55.458894 125334245051520 checkpointer.py:290] Finished synchronous save in 29.10 seconds to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/5 I0415 23:56:55.459097 125334245051520 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:56:55.459141 125334245051520 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=5] CheckpointManager Save Finalize is done on all hosts. I0415 23:56:55.459166 125334245051520 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=5] Finished synchronous save. I0415 23:56:55.459204 125334245051520 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776297386.3623657, 'wait_for_prev_duration_secs': 4.839897155761719e-05, 'time_between_consecutive_saves_sec': 0.7926878929138184, 'checkpointer_blocking_start_time': 1776297386.3625586, 'checkpointer_blocking_duration_secs': 29.096383333206177, 'get_old_steps_start_time': 1776297415.4589803, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776297386.362302, 'checkpoint_manager_blocking_duration_secs': 29.096877813339233} I0415 23:56:55.459290 125334245051520 checkpointing.py:410] Saved a checkpoint at step 5. I0415 23:56:55.460240 125334245051520 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 114.734, Tokens/s/device: 17293.938, total_weights: 16384, loss: 8.327 I0415 23:56:55.471230 125334245051520 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.492, Tokens/s/device: 17408.157, total_weights: 16384, loss: 7.367 I0415 23:56:55.583756 125334245051520 metric_logger.py:185] completed step: 6, seconds: 29.336, TFLOP/s/device: 0.463, Tokens/s/device: 69.812, total_weights: 16384, loss: 6.459 I0415 23:56:55.702500 125334245051520 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1620.023, Tokens/s/device: 244187.433, total_weights: 16384, loss: 5.622 I0415 23:56:55.820595 125334245051520 metric_logger.py:185] completed step: 8, seconds: 0.114, TFLOP/s/device: 119.467, Tokens/s/device: 18007.403, total_weights: 16384, loss: 4.864 I0415 23:56:55.938397 125334245051520 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0415 23:56:55.939214 125334245051520 checkpointing.py:776] Waited 0.000835418701171875 seconds for step 9 to finish before starting checkpointing. I0415 23:56:55.939444 125334245051520 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:56:55.939587 125334245051520 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0415 23:56:55.939640 125334245051520 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9. I0415 23:56:56.142170 125334245051520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9 W0415 23:56:56.971451 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:56:57.059130 125226127984192 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9/items I0415 23:56:57.607475 125334245051520 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:56:57.607578 125334245051520 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 23:57:01.776632 125334245051520 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.169507s I0415 23:57:01.780796 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.894 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0415 23:57:01.780854 125334245051520 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.264248s (batch_requests_ready=0.087635s, total_serialization_initiated=4.172559s, others=0.004055s) I0415 23:57:01.780930 125334245051520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.809795s (all_items=0.000025s, per_item={'items': '0.00002527'}, temp_paths=4.809770) W0415 23:57:01.780961 125334245051520 future.py:328] [process=0][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0415 23:57:02.526533 125225266054720 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_14_async_ckpt_false/checkpoints/9/items/array_metadatas/process_0 I0415 23:57:33.462190 125226784392768 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.681006s (commit=31.267071s, array_metadata_write=0.413934s) I0415 23:57:33.463484 125334245051520 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 351.541 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 35 seconds) (per-host) I0415 23:57:34.193776 125334245051520 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:57:34.584347 125334245051520 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:57:34.585071 125334245051520 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.543827s. 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_14_async_ckpt_false/checkpoints/9/items I0415 23:57:34.585723 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9/items I0415 23:57:34.811084 125334245051520 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9 I0415 23:57:35.462679 125334245051520 atomicity.py:794] [process=0][thread=MainThread] 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_14_async_ckpt_false/checkpoints/9`. I0415 23:57:35.463346 125334245051520 checkpointer.py:290] Finished synchronous save in 39.52 seconds to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_14_async_ckpt_false/checkpoints/9 I0415 23:57:35.463509 125334245051520 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:57:35.463551 125334245051520 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=9] CheckpointManager Save Finalize is done on all hosts. I0415 23:57:35.463577 125334245051520 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=9] Finished synchronous save. I0415 23:57:35.463613 125334245051520 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_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776297415.9394238, 'wait_for_prev_duration_secs': 4.482269287109375e-05, 'time_between_consecutive_saves_sec': 0.480266809463501, 'checkpointer_blocking_start_time': 1776297415.9396071, 'checkpointer_blocking_duration_secs': 39.52379560470581, 'get_old_steps_start_time': 1776297455.463429, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776297415.9393919, 'checkpoint_manager_blocking_duration_secs': 39.52419924736023} I0415 23:57:35.463722 125334245051520 checkpointing.py:410] Saved a checkpoint at step 9. I0415 23:57:35.463751 125334245051520 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:57:35.463775 125334245051520 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:57:35.464586 125334245051520 metric_logger.py:185] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.617, Tokens/s/device: 17276.286, total_weights: 16384, loss: 4.189 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops