2026-04-16 18:46:00.958004: 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 18:46:01.075419 130844700658816 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 18:46:53.755958 130844700658816 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 18:46:53.756076 130844700658816 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 18:46:53.756114 130844700658816 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 18:46:53.756141 130844700658816 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 18:46:53.756236 130844700658816 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 18:46:53.850945 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:46:54.255067 130844700658816 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:46:54.255234 130844700658816 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:46:54.255461 130844700658816 checkpointing.py:688] Setting up checkpoint logger... I0416 18:46:54.255500 130844700658816 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 18:46:54.255539 130844700658816 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 18:46:54.256156 130844700658816 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 0x770010ff8c80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 18:46:56.603040 130844700658816 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 18:46:56.603251 130844700658816 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 0x76f99193de20>}, handler_registry=None I0416 18:46:56.603525 130844700658816 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76f99193de20>` 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 18:46:56.603564 130844700658816 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76f99193e360>` 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 18:46:56.603595 130844700658816 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 0x76f99193de20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76f99193de20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76f99193e360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76f99193e360>}). I0416 18:46:56.604130 130844700658816 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 18:46:56.604179 130844700658816 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x76f99192dee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 18:46:56.729533 130844700658816 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints I0416 18:46:56.729758 130844700658816 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x76f99193c080> I0416 18:46:56.729835 130844700658816 checkpointing.py:302] Checkpoint manager created! I0416 18:46:56.813301 130844700658816 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0416 18:46:56.816474 130844700658816 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0416 18:46:56.872657 130844700658816 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0416 18:46:56.906660 130844700658816 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:46:56.906785 130844700658816 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:46:57.911747 130844700658816 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 18:46:57.911841 130844700658816 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:46:58.014890 130844700658816 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:46:58.014966 130844700658816 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:46:58.029983 130844700658816 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:46:58.030030 130844700658816 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:46:58.051315 130844700658816 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 18:46:58.051371 130844700658816 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:46:58.066579 130844700658816 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 18:46:58.066635 130844700658816 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:46:58.081758 130844700658816 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 18:46:58.081809 130844700658816 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:46:58.105393 130844700658816 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 18:46:58.105448 130844700658816 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:46:58.124039 130844700658816 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 18:46:58.124103 130844700658816 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 18:46:58.320273 130844700658816 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 18:46:58.320366 130844700658816 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 18:46:58.371197 4000111 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 18:46:58.762546 130844700658816 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 18:46:58.762652 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 18:46:58.762692 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 18:46:58.762736 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0416 18:46:58.762774 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 18:46:58.762807 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 18:46:58.762852 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0416 18:46:58.762891 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0416 18:46:58.762919 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0416 18:46:58.762945 130844700658816 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0416 18:46:58.762981 130844700658816 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 18:46:58.763014 130844700658816 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 18:46:59.195879 130844700658816 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:46:59.195962 130844700658816 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 18:46:59.210353 130844700658816 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:46:59.210402 130844700658816 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 18:46:59.503138 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:46:59.621177 130844700658816 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 18:46:59.621649 130844700658816 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 18:46:59.621788 130844700658816 metric_logger.py:289] number parameters: 1.104 billion W0416 18:47:02.758635 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:47:02.893185 130844700658816 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 18:47:03.013360 130844700658816 checkpointing.py:798] Waited 0.12016010284423828 seconds for step 0 to finish before starting checkpointing. I0416 18:47:03.013787 130844700658816 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:47:03.013935 130844700658816 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 18:47:03.014312 130844700658816 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0. I0416 18:47:03.103820 130844700658816 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 18:47:03.185678 130731453974080 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0 I0416 18:47:03.199525 130844700658816 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:47:03.199621 130844700658816 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 18:47:03.414152 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.427100 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.440076 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.456852 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.464448 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.471119 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.478224 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.482601 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:47:03.830914 130844700658816 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.631802s I0416 18:47:03.831330 130844700658816 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 16.992 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 726 milliseconds) (per-host) I0416 18:47:03.831381 130844700658816 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.726349s (batch_requests_ready=0.090759s, total_serialization_initiated=0.635273s, others=0.000317s) I0416 18:47:03.831438 130844700658816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.727150s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.727131) I0416 18:47:03.832077 130731378476608 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:47:03.832168 130844700658816 async_checkpointer.py:561] Finished blocking save. Time taken: 0.818193s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0. I0416 18:47:03.832339 130844700658816 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:47:03.832559 130844700658816 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776365223.013774, 'wait_for_prev_duration_secs': 4.1961669921875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776365223.0139568, 'checkpointer_blocking_duration_secs': 0.8183145523071289, 'get_old_steps_start_time': 1776365223.8322878, 'get_old_steps_duration_secs': 1.8835067749023438e-05, 'checkpoint_manager_blocking_start_time': 1776365223.013693, 'checkpoint_manager_blocking_duration_secs': 0.8188421726226807} I0416 18:47:03.832626 130844700658816 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 18:47:03.832662 130844700658816 max_utils.py:750] Memstats: After params initialized: I0416 18:47:03.832743 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 18:47:03.832769 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 18:47:03.832788 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 18:47:03.832806 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 18:47:03.832823 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 18:47:03.832839 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 18:47:03.832855 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 18:47:03.832871 130844700658816 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 18:47:03.838722 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.847497 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:47:03.852357 4000111 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:47:03.912623 130731443488320 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/items I0416 18:47:04.174490 130844700658816 metric_logger.py:185] completed step: 0, seconds: 3.271, TFLOP/s/device: 4.154, Tokens/s/device: 626.123, total_weights: 13328, loss: 10.828 I0416 18:47:04.175353 130844700658816 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/tensorboard/' I0416 18:47:04.280840 130731399448128 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365223819450054, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776365223819450054, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0416 18:47:04.281121 130731367990848 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:47:04.290488 130844700658816 metric_logger.py:185] completed step: 1, seconds: 1.279, TFLOP/s/device: 10.621, Tokens/s/device: 1600.849, total_weights: 12332, loss: 10.828 I0416 18:47:04.408464 130844700658816 metric_logger.py:185] completed step: 2, seconds: 0.009, TFLOP/s/device: 1509.178, Tokens/s/device: 227479.729, total_weights: 15161, loss: 9.823 I0416 18:47:04.414948 4002689 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 18:47:04.415008 4002689 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 18:47:04.525875 130844700658816 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.025, Tokens/s/device: 17488.579, total_weights: 13327, loss: 9.405 I0416 18:47:05.167839 130731453974080 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/items/array_metadatas/process_0 I0416 18:47:24.706220 130844700658816 metric_logger.py:185] completed step: 4, seconds: 0.139, TFLOP/s/device: 97.864, Tokens/s/device: 14751.111, total_weights: 11939, loss: 9.045 I0416 18:47:24.715029 130844700658816 metric_logger.py:185] completed step: 5, seconds: 0.096, TFLOP/s/device: 140.821, Tokens/s/device: 21226.097, total_weights: 15502, loss: 8.953 I0416 18:47:24.831898 130844700658816 metric_logger.py:185] completed step: 6, seconds: 20.180, TFLOP/s/device: 0.673, Tokens/s/device: 101.486, total_weights: 13864, loss: 8.775 I0416 18:47:24.949379 130844700658816 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2340.591, Tokens/s/device: 352799.311, total_weights: 12988, loss: 8.688 I0416 18:47:25.066754 130844700658816 metric_logger.py:185] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.282, Tokens/s/device: 17376.548, total_weights: 13820, loss: 8.740 I0416 18:47:25.183765 130844700658816 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 18:47:25.184549 130844700658816 checkpointing.py:798] Waited 0.0008108615875244141 seconds for step 9 to finish before starting checkpointing. I0416 18:47:25.184825 130844700658816 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:47:40.494878 130731388962368 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.079622s (commit=35.607414s, array_metadata_write=0.472208s) I0416 18:47:40.495943 130731378476608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 337.965 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0416 18:47:40.496055 130731378476608 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.663830s. I0416 18:47:40.740982 130731378476608 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365223819450054, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0416 18:47:40.922737 130731378476608 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 18:47:41.107987 130731399448128 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365223819450054, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0416 18:47:41.359286 130731378476608 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:47:41.360061 130731378476608 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.579077s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/items I0416 18:47:41.360703 130731378476608 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0/items I0416 18:47:41.601725 130731378476608 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0 I0416 18:47:42.256955 130731378476608 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0`. I0416 18:47:42.257615 130731378476608 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.243646s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/0 I0416 18:47:42.257692 130731378476608 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.425481s. I0416 18:47:42.257881 130731367990848 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:47:42.257987 130731367990848 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:47:42.258056 130731367990848 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:47:42.258102 130731367990848 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 18:47:42.258221 130844700658816 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 18:47:42.258341 130844700658816 checkpoint_manager.py:1441] Waiting for previous save to complete took 17.073525 seconds. If this number is high, consider checkpointing less frequently. I0416 18:47:42.259176 130844700658816 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 18:47:42.259455 130844700658816 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9. I0416 18:47:42.434057 130844700658816 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:47:42.434197 130731367990848 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9 I0416 18:47:42.434276 130844700658816 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 18:47:43.156070 130731244258880 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9/items I0416 18:47:51.737135 130844700658816 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.303806s I0416 18:47:51.737628 130844700658816 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.314 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0416 18:47:51.737765 130844700658816 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.394127s (batch_requests_ready=0.086738s, total_serialization_initiated=9.307001s, others=0.000388s) I0416 18:47:51.737846 130844700658816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.394641s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=9.394621) I0416 18:47:51.743582 130731344922176 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:47:51.743712 130844700658816 async_checkpointer.py:561] Finished blocking save. Time taken: 9.484481s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9. I0416 18:47:51.744038 130844700658816 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:47:51.744333 130731378476608 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:47:51.744454 130844700658816 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776365245.1847868, 'wait_for_prev_duration_secs': 17.073525428771973, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776365262.2592022, 'checkpointer_blocking_duration_secs': 9.484662532806396, 'get_old_steps_start_time': 1776365271.7438912, 'get_old_steps_duration_secs': 8.320808410644531e-05, 'checkpoint_manager_blocking_start_time': 1776365245.184737, 'checkpoint_manager_blocking_duration_secs': 26.559688568115234} I0416 18:47:51.744577 130844700658816 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 18:47:51.744613 130844700658816 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:47:52.497942 130731453974080 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9/items/array_metadatas/process_0 I0416 18:48:23.055610 130731244258880 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.313039s (commit=30.852877s, array_metadata_write=0.460162s) I0416 18:48:23.056781 130731344922176 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 310.386 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0416 18:48:23.056901 130731344922176 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.313128s. I0416 18:48:23.511426 130731344922176 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 18:48:23.985684 130731344922176 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:48:23.985858 130731344922176 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.627012s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9/items I0416 18:48:23.986430 130731344922176 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9/items I0416 18:48:24.240663 130731344922176 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9 I0416 18:48:24.942965 130731344922176 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9`. I0416 18:48:24.943608 130731344922176 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.684393s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_true/checkpoints/9 I0416 18:48:24.943688 130731344922176 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.199917s. I0416 18:48:24.943884 130731378476608 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:48:24.943989 130731378476608 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:48:24.944072 130731378476608 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:48:24.944118 130731378476608 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 18:48:24.944343 130844700658816 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 18:48:24.944952 130844700658816 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:48:24.946252 130844700658816 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 116.009, Tokens/s/device: 17486.189, total_weights: 12300, loss: 8.710 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops