2026-04-15 21:29:21.214160: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)
I0415 21:29:21.331459 125729446800512 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0415 21:30:05.490108 125729446800512 max_utils.py:800] System Information: Jax Version: 0.9.2
I0415 21:30:05.490218 125729446800512 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0415 21:30:05.490254 125729446800512 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894
I0415 21:30:05.490279 125729446800512 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0415 21:30:05.490300 125729446800512 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0415 21:30:05.490374 125729446800512 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0415 21:30:05.583254 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:05.987076 125729446800512 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0415 21:30:05.987411 125729446800512 checkpointing.py:677] Setting up checkpoint logger...
I0415 21:30:05.987454 125729446800512 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0415 21:30:05.987520 125729446800512 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0415 21:30:05.988192 125729446800512 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 0x7259141d2ff0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0415 21:30:08.424987 125729446800512 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0415 21:30:08.425228 125729446800512 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 0x725298763f80>}, handler_registry=None
I0415 21:30:08.425550 125729446800512 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x725298763f80>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`.
I0415 21:30:08.425594 125729446800512 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725297f38e60>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`.
I0415 21:30:08.425630 125729446800512 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 0x725298763f80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x725298763f80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725297f38e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725297f38e60>}).
I0415 21:30:08.425958 125729446800512 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0415 21:30:08.426015 125729446800512 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x725297f2d300> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0415 21:30:08.551896 125729446800512 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints
I0415 21:30:08.552160 125729446800512 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7254595f8440>
I0415 21:30:08.552250 125729446800512 checkpointing.py:301] Checkpoint manager created!
W0415 21:30:08.568701 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:08.747707 125729446800512 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0415 21:30:08.747807 125729446800512 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:30:08.838558 125729446800512 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:30:08.838642 125729446800512 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:30:08.852932 125729446800512 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:30:08.852980 125729446800512 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:30:08.873632 125729446800512 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:30:08.873687 125729446800512 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:30:08.888061 125729446800512 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:30:08.888107 125729446800512 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:30:08.902342 125729446800512 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:30:08.902390 125729446800512 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:30:08.924152 125729446800512 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0415 21:30:08.924209 125729446800512 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:30:08.942943 125729446800512 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0415 21:30:08.942997 125729446800512 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0415 21:30:09.127203 125729446800512 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0415 21:30:09.127299 125729446800512 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
W0415 21:30:09.176931 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
[DECOUPLED NO-OP] gcs_storage: using stubs.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] workload_monitor: using stub.
[DECOUPLED NO-OP] vertex_tensorboard: using stub.
fsdp: 8
I0415 21:30:09.560540 125729446800512 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0415 21:30:09.560654 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0415 21:30:09.560697 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0415 21:30:09.560742 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0415 21:30:09.560782 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0415 21:30:09.560814 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0415 21:30:09.560854 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/key/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0415 21:30:09.560892 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/out/kernel
Shape: float32[16,16,128,2048]
Logical: P('heads', 'layers', 'kv', 'embed')
Physical: (None, None, None, 'fsdp')
I0415 21:30:09.560921 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/query/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'q_heads', 'kv')
Physical: ('fsdp', None, None, None)
I0415 21:30:09.560946 125729446800512 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/value/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0415 21:30:09.560980 125729446800512 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0415 21:30:09.561012 125729446800512 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0415 21:30:09.984763 125729446800512 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:30:09.984846 125729446800512 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0415 21:30:09.998505 125729446800512 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:30:09.998557 125729446800512 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0415 21:30:10.643993 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:10.756000 125729446800512 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0415 21:30:10.756474 125729446800512 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0415 21:30:10.756619 125729446800512 metric_logger.py:289] number parameters: 1.104 billion
W0415 21:30:11.589125 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:11.681301 125729446800512 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0415 21:30:11.798665 125729446800512 checkpointing.py:776] Waited 0.11734700202941895 seconds for step 0 to finish before starting checkpointing.
I0415 21:30:11.799120 125729446800512 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:30:11.799270 125729446800512 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0415 21:30:11.799634 125729446800512 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0.
I0415 21:30:11.878651 125729446800512 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0415 21:30:11.967921 125621961229888 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0
I0415 21:30:11.981352 125729446800512 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:30:11.981437 125729446800512 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
W0415 21:30:12.197700 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.208376 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.222622 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.234449 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.244070 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.250140 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.255201 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.259581 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:12.516449 125729446800512 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.535509s
I0415 21:30:12.516889 125729446800512 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 19.378 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 636 milliseconds) (per-host)
I0415 21:30:12.516944 125729446800512 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.636932s (batch_requests_ready=0.097200s, total_serialization_initiated=0.539385s, others=0.000347s)
I0415 21:30:12.517004 125729446800512 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.637892s (all_items=0.000023s, per_item={'items': '0.00002313'}, temp_paths=0.637869)
I0415 21:30:12.517779 125621894121024 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:30:12.517875 125729446800512 async_checkpointer.py:561] Finished blocking save. Time taken: 0.718567s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0.
I0415 21:30:12.518058 125729446800512 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:30:12.518298 125621971715648 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:30:12.518406 125729446800512 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288611.799107, 'wait_for_prev_duration_secs': 4.1484832763671875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288611.799291, 'checkpointer_blocking_duration_secs': 0.7186846733093262, 'get_old_steps_start_time': 1776288612.5179937, 'get_old_steps_duration_secs': 1.8835067749023438e-05, 'checkpoint_manager_blocking_start_time': 1776288611.7989905, 'checkpoint_manager_blocking_duration_secs': 0.7193868160247803}
I0415 21:30:12.518516 125729446800512 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0415 21:30:12.518554 125729446800512 max_utils.py:750]
Memstats: After params initialized:
I0415 21:30:12.518631 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0415 21:30:12.518658 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0415 21:30:12.518678 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0415 21:30:12.518696 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0415 21:30:12.518713 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0415 21:30:12.518729 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0415 21:30:12.518746 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0415 21:30:12.518763 125729446800512 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0415 21:30:12.524495 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.530261 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:30:12.534988 3246383 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:30:12.671782 125621950744128 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/items
I0415 21:30:12.731120 125729446800512 metric_logger.py:185] completed step: 0, seconds: 0.924, TFLOP/s/device: 14.702, Tokens/s/device: 2215.995, total_weights: 16384, loss: 10.874
I0415 21:30:12.732007 125729446800512 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/tensorboard/'
I0415 21:30:12.808322 125621875246656 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288612567693172, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776288612567693172, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:30:12.847290 125729446800512 metric_logger.py:185] completed step: 1, seconds: 1.048, TFLOP/s/device: 12.964, Tokens/s/device: 1954.055, total_weights: 16384, loss: 10.874
I0415 21:30:12.965946 125729446800512 metric_logger.py:185] completed step: 2, seconds: 0.005, TFLOP/s/device: 2484.390, Tokens/s/device: 374474.310, total_weights: 16384, loss: 10.276
I0415 21:30:13.084130 125729446800512 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.389, Tokens/s/device: 17543.409, total_weights: 16384, loss: 9.282
I0415 21:30:13.088513 125729446800512 checkpointing.py:772] Waiting for step 5 to finish before checkpoint...
I0415 21:30:13.179743 3248545 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0415 21:30:13.179800 3248545 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0415 21:30:13.320623 125729446800512 checkpointing.py:776] Waited 0.23207426071166992 seconds for step 5 to finish before starting checkpointing.
I0415 21:30:13.321013 125729446800512 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:30:13.930584 125621915092544 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/items/array_metadatas/process_0
I0415 21:30:34.727229 125621904606784 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 21.545330s (commit=21.098925s, array_metadata_write=0.446405s)
I0415 21:30:34.728307 125621894121024 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 553.075 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 22 seconds) (per-host)
I0415 21:30:34.728407 125621894121024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 22.210487s.
I0415 21:30:34.965125 125621894121024 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288612567693172, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:30:35.151805 125621894121024 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0415 21:30:35.352520 125621875246656 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288612567693172, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:30:35.606462 125621894121024 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:30:35.606646 125621894121024 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.598438s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/items
I0415 21:30:35.607287 125621894121024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0/items
I0415 21:30:35.853215 125621894121024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0
I0415 21:30:36.527811 125621894121024 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0`.
I0415 21:30:36.528443 125621894121024 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 24.729140s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/0
I0415 21:30:36.528532 125621894121024 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 24.010613s.
I0415 21:30:36.528731 125621971715648 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:30:36.528843 125621971715648 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:30:36.528920 125621971715648 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:30:36.528966 125621971715648 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0415 21:30:36.529172 125729446800512 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.
W0415 21:30:36.529300 125729446800512 checkpoint_manager.py:1441] Waiting for previous save to complete took 23.208284 seconds. If this number is high, consider checkpointing less frequently.
I0415 21:30:36.530165 125729446800512 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5
I0415 21:30:36.530447 125729446800512 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5.
I0415 21:30:36.700409 125729446800512 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:30:36.700495 125729446800512 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0415 21:30:36.702703 125621971715648 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5
I0415 21:30:37.443017 125621613102656 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5/items
I0415 21:30:48.880709 125729446800512 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.180703s
I0415 21:30:48.881353 125729446800512 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.006 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host)
I0415 21:30:48.881414 125729446800512 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.270799s (batch_requests_ready=0.086529s, total_serialization_initiated=12.183786s, others=0.000483s)
I0415 21:30:48.881490 125729446800512 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.271279s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=12.271261)
I0415 21:30:48.887105 125621904606784 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:30:48.887235 125729446800512 async_checkpointer.py:561] Finished blocking save. Time taken: 12.357023s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5.
I0415 21:30:48.887406 125729446800512 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:30:48.887602 125621894121024 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:30:48.887721 125729446800512 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288613.3209915, 'wait_for_prev_duration_secs': 23.20828366279602, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288636.5301902, 'checkpointer_blocking_duration_secs': 12.357109546661377, 'get_old_steps_start_time': 1776288648.8873274, 'get_old_steps_duration_secs': 3.528594970703125e-05, 'checkpoint_manager_blocking_start_time': 1776288613.3209252, 'checkpoint_manager_blocking_duration_secs': 35.56676626205444}
I0415 21:30:48.887838 125729446800512 checkpointing.py:408] Started an asynchronous checkpoint save for step 5
I0415 21:30:48.888400 125729446800512 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.437, Tokens/s/device: 17249.221, total_weights: 16384, loss: 8.327
I0415 21:30:48.896908 125729446800512 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.080, Tokens/s/device: 17346.083, total_weights: 16384, loss: 7.367
I0415 21:30:49.012870 125729446800512 metric_logger.py:185] completed step: 6, seconds: 35.806, TFLOP/s/device: 0.379, Tokens/s/device: 57.197, total_weights: 16384, loss: 6.459
I0415 21:30:49.131630 125729446800512 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2254.377, Tokens/s/device: 339804.214, total_weights: 16384, loss: 5.622
I0415 21:30:49.250127 125729446800512 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.527, Tokens/s/device: 17564.172, total_weights: 16384, loss: 4.864
I0415 21:30:49.368191 125729446800512 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0415 21:30:49.369070 125729446800512 checkpointing.py:776] Waited 0.0008978843688964844 seconds for step 9 to finish before starting checkpointing.
I0415 21:30:49.369311 125729446800512 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:30:49.643107 125620365297216 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5/items/array_metadatas/process_0
I0415 21:31:15.889663 125621915092544 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 27.003223s (commit=26.560445s, array_metadata_write=0.442779s)
I0415 21:31:15.890830 125621904606784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 321.709 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host)
I0415 21:31:15.890933 125621904606784 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 27.003733s.
I0415 21:31:16.330206 125621904606784 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0415 21:31:16.789744 125621904606784 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:31:16.789920 125621904606784 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.594118s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5/items
I0415 21:31:16.790601 125621904606784 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5/items
I0415 21:31:17.036705 125621904606784 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5
I0415 21:31:17.730871 125621904606784 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5`.
I0415 21:31:17.731505 125621904606784 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.201300s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/5
I0415 21:31:17.731578 125621904606784 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.844379s.
I0415 21:31:17.731756 125621894121024 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:31:17.731864 125621894121024 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:31:17.731932 125621894121024 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:31:17.731972 125621894121024 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts.
I0415 21:31:17.732458 125729446800512 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5.
W0415 21:31:17.733645 125729446800512 checkpoint_manager.py:1441] Waiting for previous save to complete took 28.364344 seconds. If this number is high, consider checkpointing less frequently.
I0415 21:31:17.733860 125729446800512 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0415 21:31:17.734137 125729446800512 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9.
I0415 21:31:17.924249 125729446800512 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:31:17.924389 125621894121024 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9
I0415 21:31:17.924469 125729446800512 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0415 21:31:18.661522 125621613102656 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9/items
I0415 21:31:22.019746 125729446800512 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.096216s
I0415 21:31:22.023871 125729446800512 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.940 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host)
I0415 21:31:22.023932 125729446800512 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.197277s (batch_requests_ready=0.093574s, total_serialization_initiated=4.099679s, others=0.004025s)
I0415 21:31:22.024004 125729446800512 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.197761s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=4.197747)
I0415 21:31:22.025609 125621894121024 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:31:22.025724 125729446800512 async_checkpointer.py:561] Finished blocking save. Time taken: 4.291824s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9.
I0415 21:31:22.025845 125729446800512 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:31:22.025988 125621904606784 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:31:22.026096 125729446800512 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288649.3692834, 'wait_for_prev_duration_secs': 28.364344120025635, 'time_between_consecutive_saves_sec': 12.840301990509033, 'checkpointer_blocking_start_time': 1776288677.733883, 'checkpointer_blocking_duration_secs': 4.291895389556885, 'get_old_steps_start_time': 1776288682.0257897, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776288649.3692522, 'checkpoint_manager_blocking_duration_secs': 32.656819581985474}
I0415 21:31:22.026216 125729446800512 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0415 21:31:22.026248 125729446800512 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:31:22.784971 125621971715648 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9/items/array_metadatas/process_0
I0415 21:31:54.856553 125621613102656 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.832098s (commit=32.364790s, array_metadata_write=0.467309s)
I0415 21:31:54.857679 125621894121024 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 341.249 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host)
I0415 21:31:54.857731 125621894121024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.832039s.
I0415 21:31:55.276287 125621894121024 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0415 21:31:55.722080 125621894121024 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:31:55.722253 125621894121024 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.584547s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9/items
I0415 21:31:55.722882 125621894121024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9/items
I0415 21:31:55.969993 125621894121024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9
I0415 21:31:56.673906 125621894121024 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9`.
I0415 21:31:56.674515 125621894121024 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.940619s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9
I0415 21:31:56.674582 125621894121024 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.648889s.
I0415 21:31:56.674738 125621904606784 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:31:56.674842 125621904606784 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:31:56.674891 125621904606784 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:31:56.674932 125621904606784 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0415 21:31:56.675292 125729446800512 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.
I0415 21:31:56.675971 125729446800512 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:31:56.677259 125729446800512 metric_logger.py:185] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.399, Tokens/s/device: 17243.412, total_weights: 16384, loss: 4.189
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops