2026-04-16 03:38:17.318568: 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 03:38:17.433798 127667789679744 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 03:39:14.205353 127667789679744 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 03:39:14.205458 127667789679744 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 03:39:14.205508 127667789679744 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 03:39:14.205536 127667789679744 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 03:39:14.205564 127667789679744 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 03:39:14.205673 127667789679744 train.py:818] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 03:39:14.305224 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:14.709578 127667789679744 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 03:39:14.709755 127667789679744 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 03:39:14.710192 127667789679744 checkpointing.py:688] Setting up checkpoint logger...
I0416 03:39:14.710235 127667789679744 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 03:39:14.710276 127667789679744 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 03:39:14.711237 127667789679744 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 0x741c625f8c20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 03:39:17.046914 127667789679744 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 03:39:17.047115 127667789679744 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 0x7415e5f36a50>}, handler_registry=None
I0416 03:39:17.047412 127667789679744 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7415e5f36a50>` 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 03:39:17.047453 127667789679744 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7415e5f36ff0>` 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 03:39:17.047482 127667789679744 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 0x7415e5f36a50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7415e5f36a50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7415e5f36ff0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7415e5f36ff0>}).
I0416 03:39:17.047799 127667789679744 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 03:39:17.047849 127667789679744 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7415e5f420c0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 03:39:17.164324 127667789679744 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints
I0416 03:39:17.164548 127667789679744 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_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7415e5f36570>
I0416 03:39:17.164634 127667789679744 checkpointing.py:302] Checkpoint manager created!
W0416 03:39:17.180848 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:17.369256 127667789679744 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 03:39:17.369345 127667789679744 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 03:39:17.468256 127667789679744 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 03:39:17.468391 127667789679744 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 03:39:17.482798 127667789679744 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 03:39:17.482846 127667789679744 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 03:39:17.504927 127667789679744 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 03:39:17.504985 127667789679744 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 03:39:17.519264 127667789679744 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 03:39:17.519313 127667789679744 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 03:39:17.533591 127667789679744 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 03:39:17.533652 127667789679744 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 03:39:17.560706 127667789679744 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 03:39:17.560790 127667789679744 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 03:39:17.579129 127667789679744 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 03:39:17.579181 127667789679744 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 03:39:17.771262 127667789679744 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 03:39:17.771364 127667789679744 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0416 03:39:17.822733 3626895 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 03:39:18.220244 127667789679744 maxtext_utils.py:1790] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 03:39:18.220409 127667789679744 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 03:39:18.220451 127667789679744 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 03:39:18.220496 127667789679744 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0416 03:39:18.220541 127667789679744 maxtext_utils.py:1790] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 03:39:18.220569 127667789679744 maxtext_utils.py:1790] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 03:39:18.220610 127667789679744 maxtext_utils.py:1790] 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 03:39:18.220650 127667789679744 maxtext_utils.py:1790] 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 03:39:18.220678 127667789679744 maxtext_utils.py:1790] 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 03:39:18.220703 127667789679744 maxtext_utils.py:1790] 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 03:39:18.220738 127667789679744 maxtext_utils.py:1790] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 03:39:18.220773 127667789679744 maxtext_utils.py:1790] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 03:39:18.647482 127667789679744 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 03:39:18.647564 127667789679744 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 03:39:18.662396 127667789679744 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 03:39:18.662447 127667789679744 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 03:39:19.339428 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:19.465547 127667789679744 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 03:39:19.466023 127667789679744 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 03:39:19.466182 127667789679744 metric_logger.py:289] number parameters: 1.104 billion
W0416 03:39:20.339541 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:20.436081 127667789679744 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0416 03:39:20.553494 127667789679744 checkpointing.py:798] Waited 0.11741757392883301 seconds for step 0 to finish before starting checkpointing.
I0416 03:39:20.553941 127667789679744 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 03:39:20.554127 127667789679744 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 03:39:20.554496 127667789679744 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0.
I0416 03:39:20.644660 127667789679744 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 03:39:20.745153 127560287520320 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0
I0416 03:39:20.745821 127667789679744 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 03:39:20.745980 127667789679744 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 03:39:20.959262 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:20.967181 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:20.979658 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:20.993195 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.003981 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.010124 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.015186 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.019389 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:21.332093 127667789679744 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.587205s
I0416 03:39:21.332494 127667789679744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 17.968 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 686 milliseconds) (per-host)
I0416 03:39:21.332544 127667789679744 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.686923s (batch_requests_ready=0.095514s, total_serialization_initiated=0.591099s, others=0.000310s)
I0416 03:39:21.332608 127667789679744 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.687466s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.687446)
I0416 03:39:21.333227 127560220411456 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 03:39:21.333308 127667789679744 async_checkpointer.py:561] Finished blocking save. Time taken: 0.779141s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0.
I0416 03:39:21.333489 127667789679744 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 03:39:21.333660 127667789679744 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776310760.5539272, 'wait_for_prev_duration_secs': 4.1484832763671875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776310760.5541494, 'checkpointer_blocking_duration_secs': 0.7792623043060303, 'get_old_steps_start_time': 1776310761.333431, 'get_old_steps_duration_secs': 2.3603439331054688e-05, 'checkpoint_manager_blocking_start_time': 1776310760.5538225, 'checkpoint_manager_blocking_duration_secs': 0.7798135280609131}
I0416 03:39:21.333722 127667789679744 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0416 03:39:21.333757 127667789679744 max_utils.py:750]
Memstats: After params initialized:
I0416 03:39:21.333838 127667789679744 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0))
I0416 03:39:21.333864 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 03:39:21.333885 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 03:39:21.333904 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 03:39:21.333920 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 03:39:21.333936 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 03:39:21.333953 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 03:39:21.333968 127667789679744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 03:39:21.339238 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.344870 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:39:21.349961 3626895 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:39:21.396011 127560277034560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/items
I0416 03:39:21.547359 127560241382976 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310761305604270, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776310761305604270, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:39:21.547592 127560298006080 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 03:39:21.714939 127667789679744 metric_logger.py:185] completed step: 0, seconds: 0.969, TFLOP/s/device: 14.016, Tokens/s/device: 2112.612, total_weights: 16384, loss: 10.874
I0416 03:39:21.715843 127667789679744 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/tensorboard/'
I0416 03:39:21.831007 127667789679744 metric_logger.py:185] completed step: 1, seconds: 1.277, TFLOP/s/device: 10.639, Tokens/s/device: 1603.575, total_weights: 16384, loss: 10.874
I0416 03:39:21.907410 3629171 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 03:39:21.907470 3629171 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 03:39:21.949923 127667789679744 metric_logger.py:185] completed step: 2, seconds: 0.006, TFLOP/s/device: 2171.509, Tokens/s/device: 327313.409, total_weights: 16384, loss: 9.775
I0416 03:39:22.068758 127667789679744 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.273, Tokens/s/device: 17525.994, total_weights: 16384, loss: 8.838
I0416 03:39:22.854264 127560287520320 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/items/array_metadatas/process_0
I0416 03:39:41.620663 127667789679744 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.065, Tokens/s/device: 17193.180, total_weights: 16384, loss: 7.965
I0416 03:39:41.627209 127667789679744 metric_logger.py:185] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.386, Tokens/s/device: 17241.524, total_weights: 16384, loss: 7.233
I0416 03:39:41.656888 127560230897216 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 19.746172s (commit=19.280092s, array_metadata_write=0.466080s)
I0416 03:39:41.658036 127560220411456 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 601.399 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 21 seconds) (per-host)
I0416 03:39:41.658159 127560220411456 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.324805s.
I0416 03:39:41.744451 127667789679744 metric_logger.py:185] completed step: 6, seconds: 19.552, TFLOP/s/device: 0.695, Tokens/s/device: 104.746, total_weights: 16384, loss: 6.676
I0416 03:39:41.863022 127667789679744 metric_logger.py:185] completed step: 7, seconds: 0.004, TFLOP/s/device: 3127.072, Tokens/s/device: 471346.375, total_weights: 16384, loss: 6.293
I0416 03:39:41.879746 127560220411456 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310761305604270, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:39:41.981548 127667789679744 metric_logger.py:185] completed step: 8, seconds: 0.118, TFLOP/s/device: 114.845, Tokens/s/device: 17310.748, total_weights: 16384, loss: 6.055
I0416 03:39:42.062803 127560220411456 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 03:39:42.100292 127667789679744 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0416 03:39:42.101156 127667789679744 checkpointing.py:798] Waited 0.0008940696716308594 seconds for step 9 to finish before starting checkpointing.
I0416 03:39:42.101410 127667789679744 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 03:39:42.262987 127560241382976 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310761305604270, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:39:42.476414 127560220411456 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 03:39:42.477082 127560220411456 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.553999s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/items
I0416 03:39:42.477754 127560220411456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0/items
I0416 03:39:42.698334 127560220411456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0
I0416 03:39:43.380218 127560220411456 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0`.
I0416 03:39:43.380852 127560220411456 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 22.826690s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/0
I0416 03:39:43.380928 127560220411456 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 22.047577s.
I0416 03:39:43.381121 127560298006080 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 03:39:43.381230 127560298006080 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 03:39:43.381288 127560298006080 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 03:39:43.381334 127560298006080 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 03:39:43.381427 127667789679744 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 03:39:43.381490 127667789679744 checkpoint_manager.py:1441] Waiting for previous save to complete took 1.280090 seconds. If this number is high, consider checkpointing less frequently.
I0416 03:39:43.382253 127667789679744 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 03:39:43.382532 127667789679744 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9.
I0416 03:39:43.571448 127560298006080 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9
I0416 03:39:43.581457 127667789679744 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 03:39:43.581549 127667789679744 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 03:39:44.252980 127560077805120 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9/items
I0416 03:39:54.798470 127667789679744 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.217420s
I0416 03:39:54.802700 127667789679744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.090 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host)
I0416 03:39:54.802765 127667789679744 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.325191s (batch_requests_ready=0.100425s, total_serialization_initiated=11.220637s, others=0.004129s)
I0416 03:39:54.802837 127667789679744 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.325678s (all_items=0.000021s, per_item={'items': '0.00002098'}, temp_paths=11.325657)
I0416 03:39:54.804254 127560287520320 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 03:39:54.804335 127667789679744 async_checkpointer.py:561] Finished blocking save. Time taken: 11.422040s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9.
I0416 03:39:54.804521 127667789679744 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 03:39:54.804694 127560220411456 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 03:39:54.804804 127667789679744 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776310782.1013782, 'wait_for_prev_duration_secs': 1.28009033203125, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776310783.382276, 'checkpointer_blocking_duration_secs': 11.422163009643555, 'get_old_steps_start_time': 1776310794.8044538, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1776310782.1013415, 'checkpoint_manager_blocking_duration_secs': 12.703437566757202}
I0416 03:39:54.804914 127667789679744 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0416 03:39:54.804948 127667789679744 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 03:39:55.574567 127559490602560 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9/items/array_metadatas/process_0
I0416 03:40:22.534645 127559406716480 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 27.731269s (commit=27.305471s, array_metadata_write=0.425799s)
I0416 03:40:22.535583 127560287520320 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 323.539 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host)
I0416 03:40:22.535640 127560287520320 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 27.731262s.
I0416 03:40:23.003648 127560287520320 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 03:40:23.406204 127560287520320 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 03:40:23.406381 127560287520320 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.569818s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9/items
I0416 03:40:23.406970 127560287520320 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9/items
I0416 03:40:23.647832 127560287520320 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9
I0416 03:40:24.341337 127560287520320 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9`.
I0416 03:40:24.341991 127560287520320 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.959702s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_02_synthetic/checkpoints/9
I0416 03:40:24.342076 127560287520320 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 29.537699s.
I0416 03:40:24.342238 127560220411456 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 03:40:24.342344 127560220411456 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 03:40:24.342404 127560220411456 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 03:40:24.342449 127560220411456 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 03:40:24.342693 127667789679744 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 03:40:24.343160 127667789679744 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 03:40:24.343839 127667789679744 metric_logger.py:185] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.647, Tokens/s/device: 17280.805, total_weights: 16384, loss: 5.915
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops