MaxView

2026-04-23 02:57:35.896447: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1776913055.910006 2350703 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1776913055.913997 2350703 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1776913055.926492 2350703 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776913055.926506 2350703 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776913055.926508 2350703 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776913055.926509 2350703 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
`rope_scaling`'s factor field must be a float >= 1, got 40
`rope_scaling`'s beta_fast field must be a float, got 32
`rope_scaling`'s beta_slow field must be a float, got 1
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
2026-04-23 02:57:39.598625: 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)
I0423 02:57:39.736911 126415161100096 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0423 02:58:36.347659 126415161100096 max_utils.py:800] System Information: Jax Version: 0.8.1
I0423 02:58:36.347911 126415161100096 max_utils.py:801] System Information: Jaxlib Version: 0.8.1
I0423 02:58:36.347958 126415161100096 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709
I0423 02:58:36.347987 126415161100096 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system
I0423 02:58:36.348013 126415161100096 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 02:58:36.348186 126415161100096 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
I0423 02:58:37.008850 126415161100096 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 02:58:37.009130 126415161100096 checkpointing.py:677] Setting up checkpoint logger...
I0423 02:58:37.009174 126415161100096 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0423 02:58:37.009209 126415161100096 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0423 02:58:37.009624 126415161100096 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 0x72f9093a9730>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 02:58:39.411940 126415161100096 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0423 02:58:39.412313 126415161100096 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 0x72f735bc9cd0>}, handler_registry=None
I0423 02:58:39.412534 126415161100096 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x72f735bc9cd0>` 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`.
I0423 02:58:39.412576 126415161100096 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72f735bca0f0>` 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`.
I0423 02:58:39.412605 126415161100096 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 0x72f735bc9cd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x72f735bc9cd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72f735bca0f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72f735bca0f0>}).
I0423 02:58:39.412878 126415161100096 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0423 02:58:39.412930 126415161100096 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x72f735be04a0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0423 02:58:39.542468 126415161100096 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints
I0423 02:58:39.542684 126415161100096 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_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x72f735bc96d0>
I0423 02:58:39.542960 126415161100096 checkpointing.py:301] Checkpoint manager created!
I0423 02:58:40.482839 126415161100096 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 02:58:40.482934 126415161100096 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 02:58:40.763900 126415161100096 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 02:58:40.763979 126415161100096 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 02:58:40.776156 126415161100096 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 02:58:40.776204 126415161100096 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0423 02:58:40.795191 126415161100096 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 02:58:40.795246 126415161100096 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 02:58:40.807571 126415161100096 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 02:58:40.807621 126415161100096 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 02:58:40.819716 126415161100096 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 02:58:40.819765 126415161100096 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 02:58:40.839152 126415161100096 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 02:58:40.839209 126415161100096 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0423 02:58:40.854573 126415161100096 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 02:58:40.854625 126415161100096 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0423 02:58:41.004991 126415161100096 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 02:58:41.005095 126415161100096 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
[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
I0423 02:58:41.636674 126415161100096 maxtext_utils.py:1707]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   PartitionSpec('norm',)
    Physical:  (None,)
I0423 02:58:41.636784 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0423 02:58:41.636828 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0423 02:58:41.636875 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   PartitionSpec('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0423 02:58:41.636916 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0423 02:58:41.636945 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0423 02:58:41.636988 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0423 02:58:41.637029 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   PartitionSpec('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0423 02:58:41.637072 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0423 02:58:41.637102 126415161100096 maxtext_utils.py:1707]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0423 02:58:41.637137 126415161100096 maxtext_utils.py:1707]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   PartitionSpec('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0423 02:58:41.637173 126415161100096 maxtext_utils.py:1707]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   PartitionSpec('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0423 02:58:41.654017 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown.
I0423 02:58:41.654079 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,).
I0423 02:58:41.665064 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown.
I0423 02:58:41.665108 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None).
I0423 02:58:41.686966 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown.
I0423 02:58:41.687020 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp').
I0423 02:58:41.698036 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown.
I0423 02:58:41.698092 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None).
I0423 02:58:41.719923 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown.
I0423 02:58:41.719975 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None).
I0423 02:58:41.730974 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown.
I0423 02:58:41.731020 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp').
I0423 02:58:41.763697 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown.
I0423 02:58:41.763756 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None).
I0423 02:58:41.774715 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown.
I0423 02:58:41.774758 126415161100096 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp').
I0423 02:58:42.302123 126415161100096 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0423 02:58:42.302202 126415161100096 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0423 02:58:42.314218 126415161100096 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0423 02:58:42.314269 126415161100096 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
I0423 02:58:43.310529 126415161100096 max_utils.py:791] Total memory size: 4.5 GB, Output size: 1.5 GB, Temp size: 2.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0423 02:58:43.311094 126415161100096 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0423 02:58:43.311249 126415161100096 metric_logger.py:301] number parameters: 1.104 billion
I0423 02:58:45.921437 126415161100096 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0423 02:58:46.343466 126415161100096 checkpointing.py:776] Waited 0.42201876640319824 seconds for step 0 to finish before starting checkpointing.
I0423 02:58:46.344018 126415161100096 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 02:58:46.344185 126415161100096 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0423 02:58:46.344547 126415161100096 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0.
I0423 02:58:46.428919 126415161100096 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0423 02:58:46.518795 126383772665408 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0
I0423 02:58:46.523483 126415161100096 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0423 02:58:46.523571 126415161100096 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
I0423 02:58:47.299100 126383762179648 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/items
I0423 02:58:47.452820 126383730722368 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776913127219267975, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776913127219267975, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0423 02:58:47.540140 126415161100096 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.017025s
I0423 02:58:47.540555 126415161100096 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 11.114 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host)
I0423 02:58:47.540611 126415161100096 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.110491s (batch_requests_ready=0.089679s, total_serialization_initiated=1.020488s, others=0.000324s)
I0423 02:58:47.540674 126415161100096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.111304s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=1.111286)
I0423 02:58:47.541393 126383535687232 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0423 02:58:47.541486 126415161100096 async_checkpointer.py:561] Finished blocking save. Time taken: 1.197261s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0.
I0423 02:58:47.541690 126415161100096 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 02:58:47.541875 126383921563200 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0423 02:58:47.541982 126415161100096 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776913126.3440053, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776913126.3442073, 'checkpointer_blocking_duration_secs': 1.1973886489868164, 'get_old_steps_start_time': 1776913127.541622, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776913126.3439276, 'checkpoint_manager_blocking_duration_secs': 1.1980290412902832}
I0423 02:58:47.542116 126415161100096 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0423 02:58:47.542170 126415161100096 max_utils.py:750] 
Memstats: After params initialized:
I0423 02:58:47.542237 126415161100096 max_utils.py:756] 	Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0))
I0423 02:58:47.542264 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0423 02:58:47.542286 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0423 02:58:47.542304 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0423 02:58:47.542321 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0423 02:58:47.542341 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0423 02:58:47.542358 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0423 02:58:47.542380 126415161100096 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
I0423 02:58:47.901966 2352698 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0423 02:58:47.902028 2352698 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0423 02:58:47.932320 126415161100096 metric_logger.py:196] completed step: 0, seconds: 2.610, TFLOP/s/device: 20.826, Tokens/s/device: 3139.086, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648
I0423 02:58:47.933266 126415161100096 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/tensorboard/'
I0423 02:58:48.328940 126415161100096 metric_logger.py:196] completed step: 1, seconds: 1.989, TFLOP/s/device: 27.324, Tokens/s/device: 4118.549, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648
I0423 02:58:48.660001 126383720236608 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/items/array_metadatas/process_0
I0423 02:58:48.764405 126415161100096 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 2005.258, Tokens/s/device: 302254.363, total_weights: 65536, loss: 10.261, lm_loss: 10.261, perplexity: 28586.895
I0423 02:58:49.189096 126415161100096 metric_logger.py:196] completed step: 3, seconds: 0.397, TFLOP/s/device: 136.729, Tokens/s/device: 20609.324, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16845.295
I0423 02:59:22.138855 126415161100096 metric_logger.py:196] completed step: 4, seconds: 0.436, TFLOP/s/device: 124.762, Tokens/s/device: 18805.510, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10654.189
I0423 02:59:22.147230 126415161100096 metric_logger.py:196] completed step: 5, seconds: 0.426, TFLOP/s/device: 127.546, Tokens/s/device: 19225.128, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7241.146
I0423 02:59:22.571494 126415161100096 metric_logger.py:196] completed step: 6, seconds: 32.949, TFLOP/s/device: 1.649, Tokens/s/device: 248.625, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5368.870
I0423 02:59:22.989800 126415161100096 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 9729.416, Tokens/s/device: 1466523.451, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4361.125
I0423 02:59:23.414214 126415161100096 metric_logger.py:196] completed step: 8, seconds: 0.425, TFLOP/s/device: 127.823, Tokens/s/device: 19266.862, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3832.059
I0423 02:59:23.842336 126415161100096 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0423 02:59:23.843271 126415161100096 checkpointing.py:776] Waited 0.0009624958038330078 seconds for step 9 to finish before starting checkpointing.
I0423 02:59:23.843547 126415161100096 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 02:59:32.364466 126383751693888 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 44.460208s (commit=43.970188s, array_metadata_write=0.490020s)
I0423 02:59:32.365391 126383535687232 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 275.100 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host)
I0423 02:59:32.365444 126383535687232 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.823914s.
I0423 02:59:32.620813 126383535687232 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776913127219267975, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0423 02:59:32.851247 126383535687232 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0423 02:59:33.107495 126383730722368 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776913127219267975, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0423 02:59:33.253514 126383535687232 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0423 02:59:33.253689 126383535687232 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570245s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/items
I0423 02:59:33.254522 126383535687232 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0/items
I0423 02:59:33.492303 126383535687232 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0
I0423 02:59:34.246766 126383535687232 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0`.
I0423 02:59:34.247386 126383535687232 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.903166s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/0
I0423 02:59:34.247464 126383535687232 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 46.705935s.
I0423 02:59:34.247660 126383921563200 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 02:59:34.247770 126383921563200 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0423 02:59:34.247833 126383921563200 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0423 02:59:34.247877 126383921563200 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0423 02:59:34.247972 126415161100096 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.
W0423 02:59:34.248031 126415161100096 checkpoint_manager.py:1441] Waiting for previous save to complete took 10.404500 seconds. If this number is high, consider checkpointing less frequently.
I0423 02:59:34.249034 126415161100096 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0423 02:59:34.249314 126415161100096 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9.
I0423 02:59:34.425616 126415161100096 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0423 02:59:34.425706 126415161100096 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
I0423 02:59:34.427587 126383535687232 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9
I0423 02:59:35.175710 126383720236608 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9/items
I0423 02:59:39.442733 126415161100096 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.017466s
I0423 02:59:39.447057 126415161100096 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.412 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host)
I0423 02:59:39.447121 126415161100096 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.115721s (batch_requests_ready=0.091055s, total_serialization_initiated=5.020439s, others=0.004227s)
I0423 02:59:39.447201 126415161100096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.116262s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=5.116246)
I0423 02:59:39.449606 126383535687232 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0423 02:59:39.449714 126415161100096 async_checkpointer.py:561] Finished blocking save. Time taken: 5.200624s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9.
I0423 02:59:39.449901 126415161100096 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 02:59:39.450135 126383751693888 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0423 02:59:39.450248 126415161100096 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776913163.843513, 'wait_for_prev_duration_secs': 10.404499530792236, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776913174.2490714, 'checkpointer_blocking_duration_secs': 5.200751543045044, 'get_old_steps_start_time': 1776913179.4498367, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776913163.843475, 'checkpoint_manager_blocking_duration_secs': 15.606746435165405}
I0423 02:59:39.450371 126415161100096 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0423 02:59:39.450407 126415161100096 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 02:59:40.210646 126383571338816 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9/items/array_metadatas/process_0
I0423 03:00:29.143142 126383720236608 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 49.695051s (commit=49.214884s, array_metadata_write=0.480167s)
I0423 03:00:29.144303 126383535687232 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 230.544 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 54 seconds) (per-host)
I0423 03:00:29.144417 126383535687232 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.694657s.
I0423 03:00:29.609017 126383535687232 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0423 03:00:30.017462 126383535687232 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0423 03:00:30.017727 126383535687232 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.584713s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9/items
I0423 03:00:30.018396 126383535687232 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9/items
I0423 03:00:30.268272 126383535687232 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9
I0423 03:00:30.999149 126383535687232 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9`.
I0423 03:00:30.999756 126383535687232 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 56.750672s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_06_grad_accum/checkpoints/9
I0423 03:00:30.999833 126383535687232 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.550076s.
I0423 03:00:31.000066 126383751693888 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 03:00:31.000172 126383751693888 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0423 03:00:31.000237 126383751693888 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0423 03:00:31.000285 126383751693888 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0423 03:00:31.000426 126415161100096 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.
I0423 03:00:31.000555 126415161100096 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 03:00:31.001488 126415161100096 metric_logger.py:196] completed step: 9, seconds: 0.418, TFLOP/s/device: 129.919, Tokens/s/device: 19582.860, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3551.987
Per train step:
 Total TFLOPs: 54.35 
 split as 93.93% learnable weight flops and 6.07% attention flops