2026-04-16 13:52:07.223847: 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 13:52:07.341427 131281465216128 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 13:52:46.165580 131281465216128 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 13:52:46.165703 131281465216128 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 13:52:46.165741 131281465216128 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 13:52:46.165767 131281465216128 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0416 13:52:46.165793 131281465216128 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 13:52:46.165879 131281465216128 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 13:52:46.259030 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:52:46.675627 131281465216128 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 13:52:46.675958 131281465216128 checkpointing.py:677] Setting up checkpoint logger... I0416 13:52:46.675999 131281465216128 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0416 13:52:46.676041 131281465216128 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 13:52:46.677432 131281465216128 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7765c2ffcc80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 13:52:48.489234 131281465216128 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0416 13:52:48.489426 131281465216128 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 0x775f4673dbb0>}, handler_registry=None I0416 13:52:48.489686 131281465216128 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x775f4673dbb0>` 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 13:52:48.489727 131281465216128 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x775f4673e120>` 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 13:52:48.489759 131281465216128 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 0x775f4673dbb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x775f4673dbb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x775f4673e120>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x775f4673e120>}). I0416 13:52:48.490221 131281465216128 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 13:52:48.490275 131281465216128 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x775f46731260> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 13:52:48.905107 131175343457856 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776347461919231741, 'commit_timestamp_nsecs': 1776347502038015971, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0416 13:52:48.993828 131175343457856 checkpoint_manager.py:1847] Missing metrics for step 9 I0416 13:52:49.012733 131281465216128 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints I0416 13:52:49.013264 131281465216128 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_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x776107978aa0> I0416 13:52:49.013359 131281465216128 checkpointing.py:301] Checkpoint manager created! W0416 13:52:49.030655 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:52:49.221624 131281465216128 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 13:52:49.221796 131281465216128 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:52:49.317067 131281465216128 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 13:52:49.317156 131281465216128 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:52:49.331603 131281465216128 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 13:52:49.331657 131281465216128 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:52:49.352792 131281465216128 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 13:52:49.352853 131281465216128 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:52:49.366972 131281465216128 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 13:52:49.367031 131281465216128 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:52:49.385678 131281465216128 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 13:52:49.385779 131281465216128 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:52:49.407967 131281465216128 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 13:52:49.408034 131281465216128 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:52:49.426327 131281465216128 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 13:52:49.426383 131281465216128 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 13:52:49.609591 131281465216128 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0416 13:52:49.609694 131281465216128 checkpointing.py:581] restoring from this run's directory step 9 I0416 13:52:49.773334 131281465216128 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/9. I0416 13:52:50.284002 3894660 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 13:52:50.284092 3894660 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 13:52:56.270548 131281465216128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 2.000 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 13:52:56.270839 131281465216128 checkpointer.py:318] Finished restoring checkpoint in 6.58 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/9. I0416 13:52:56.270924 131281465216128 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776347569.6904366, 'checkpointer_duration_secs': 6.580448865890503, 'checkpoint_manager_start_time': 1776347569.6099143, 'checkpoint_manager_duration_secs': 6.660974502563477} [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 13:52:56.841353 131281465216128 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 13:52:56.841470 131281465216128 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) I0416 13:52:56.841511 131281465216128 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) I0416 13:52:56.841554 131281465216128 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') I0416 13:52:56.841593 131281465216128 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) I0416 13:52:56.841620 131281465216128 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) I0416 13:52:56.841659 131281465216128 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) I0416 13:52:56.841699 131281465216128 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') I0416 13:52:56.841729 131281465216128 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) I0416 13:52:56.841755 131281465216128 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) I0416 13:52:56.841795 131281465216128 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 13:52:56.841831 131281465216128 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 13:52:57.321440 131281465216128 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 13:52:57.321531 131281465216128 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 13:52:57.335193 131281465216128 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 13:52:57.335248 131281465216128 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 13:52:57.983490 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:52:58.053875 131281465216128 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 13:52:58.054569 131281465216128 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 13:52:58.054724 131281465216128 metric_logger.py:289] number parameters: 1.104 billion W0416 13:52:58.889096 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:52:58.966111 131281465216128 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0416 13:52:59.082378 131281465216128 checkpointing.py:776] Waited 0.11623334884643555 seconds for step 10 to finish before starting checkpointing. I0416 13:52:59.082852 131281465216128 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 13:52:59.083026 131281465216128 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0416 13:52:59.083362 131281465216128 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10. I0416 13:52:59.161432 131281465216128 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 13:52:59.242516 131173787371072 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10 I0416 13:52:59.251335 131281465216128 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:52:59.251434 131281465216128 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 13:52:59.493731 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.545608 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.550168 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.554397 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.558427 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.562396 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.566208 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:52:59.570149 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:52:59.984620 131173776885312 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/items I0416 13:53:00.174674 131173745428032 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776347579897195795, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776347579897195795, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0416 13:53:03.626481 131281465216128 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.375646s I0416 13:53:03.630761 131281465216128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.762 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0416 13:53:03.630822 131281465216128 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.468342s (batch_requests_ready=0.085600s, total_serialization_initiated=4.378585s, others=0.004156s) I0416 13:53:03.630895 131281465216128 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.468941s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=4.468921) I0416 13:53:03.632254 131173797856832 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:53:03.632339 131281465216128 async_checkpointer.py:561] Finished blocking save. Time taken: 4.549249s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10. I0416 13:53:03.632528 131281465216128 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:53:03.632735 131173787371072 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:53:03.632853 131281465216128 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347579.0828388, 'wait_for_prev_duration_secs': 3.933906555175781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776347579.083063, 'checkpointer_blocking_duration_secs': 4.5493810176849365, 'get_old_steps_start_time': 1776347583.6324646, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776347579.082745, 'checkpoint_manager_blocking_duration_secs': 4.550080299377441} I0416 13:53:03.632967 131281465216128 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0416 13:53:03.633006 131281465216128 max_utils.py:750] Memstats: After params initialized: I0416 13:53:03.633084 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 13:53:03.633114 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 13:53:03.633136 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 13:53:03.633154 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 13:53:03.633172 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 13:53:03.633190 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 13:53:03.633208 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 13:53:03.633225 131281465216128 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 13:53:03.639837 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:53:03.646319 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:53:03.650933 3893114 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:53:03.858814 131281465216128 metric_logger.py:185] completed step: 10, seconds: 0.911, TFLOP/s/device: 14.916, Tokens/s/device: 2248.331, total_weights: 16384, loss: 3.602 I0416 13:53:03.974840 131281465216128 metric_logger.py:185] completed step: 11, seconds: 4.891, TFLOP/s/device: 2.778, Tokens/s/device: 418.731, total_weights: 16384, loss: 3.101 I0416 13:53:04.093385 131281465216128 metric_logger.py:185] completed step: 12, seconds: 0.005, TFLOP/s/device: 2496.716, Tokens/s/device: 376332.231, total_weights: 16384, loss: 2.684 I0416 13:53:04.211864 131281465216128 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.964, Tokens/s/device: 17479.324, total_weights: 16384, loss: 2.348 I0416 13:53:04.216758 131281465216128 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0416 13:53:04.412976 131173755913792 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/items/array_metadatas/process_0 I0416 13:53:04.447859 131281465216128 checkpointing.py:776] Waited 0.23106002807617188 seconds for step 15 to finish before starting checkpointing. I0416 13:53:04.448303 131281465216128 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:53:33.083506 131173776885312 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.451967s (commit=28.999472s, array_metadata_write=0.452495s) I0416 13:53:33.084861 131173797856832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 372.521 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0416 13:53:33.084980 131173797856832 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.452679s. I0416 13:53:33.502557 131173797856832 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 13:53:33.712716 131173745428032 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776347579897195795, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0416 13:53:34.011209 131173797856832 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.641384s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/items I0416 13:53:34.012016 131173797856832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10/items I0416 13:53:34.266373 131173797856832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10 I0416 13:53:34.954278 131173797856832 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10`. I0416 13:53:34.955022 131173797856832 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.871947s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/10 I0416 13:53:34.955121 131173797856832 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.322822s. I0416 13:53:34.955348 131173787371072 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:53:34.955472 131173787371072 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:53:34.955540 131173787371072 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:53:34.955589 131173787371072 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0416 13:53:34.956067 131281465216128 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0416 13:53:34.957244 131281465216128 checkpoint_manager.py:1441] Waiting for previous save to complete took 30.508939 seconds. If this number is high, consider checkpointing less frequently. I0416 13:53:34.958581 131281465216128 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0416 13:53:34.958885 131281465216128 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15. I0416 13:53:35.136269 131281465216128 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:53:35.136370 131281465216128 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 13:53:35.142665 131173787371072 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15 I0416 13:53:35.878822 131173734942272 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15/items I0416 13:53:43.175488 131281465216128 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.039702s I0416 13:53:43.180066 131281465216128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.517 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0416 13:53:43.180199 131281465216128 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.133650s (batch_requests_ready=0.085594s, total_serialization_initiated=8.043530s, others=0.004526s) I0416 13:53:43.180280 131281465216128 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.134212s (all_items=0.000015s, per_item={'items': '0.00001550'}, temp_paths=8.134196) I0416 13:53:43.181821 131173787371072 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:53:43.181907 131281465216128 async_checkpointer.py:561] Finished blocking save. Time taken: 8.223281s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15. I0416 13:53:43.182116 131281465216128 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:53:43.182325 131175259571776 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:53:43.182431 131281465216128 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347584.4482787, 'wait_for_prev_duration_secs': 30.508938550949097, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776347614.958607, 'checkpointer_blocking_duration_secs': 8.223401308059692, 'get_old_steps_start_time': 1776347623.1820297, 'get_old_steps_duration_secs': 4.315376281738281e-05, 'checkpoint_manager_blocking_start_time': 1776347584.4482117, 'checkpoint_manager_blocking_duration_secs': 38.73419260978699} I0416 13:53:43.182553 131281465216128 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0416 13:53:43.183059 131281465216128 metric_logger.py:185] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.446, Tokens/s/device: 17250.529, total_weights: 16384, loss: 2.083 I0416 13:53:43.191838 131281465216128 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 114.956, Tokens/s/device: 17327.445, total_weights: 16384, loss: 1.880 I0416 13:53:43.307893 131281465216128 metric_logger.py:185] completed step: 16, seconds: 38.973, TFLOP/s/device: 0.349, Tokens/s/device: 52.549, total_weights: 16384, loss: 1.726 I0416 13:53:43.426073 131281465216128 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2192.180, Tokens/s/device: 330429.171, total_weights: 16384, loss: 1.610 I0416 13:53:43.544137 131281465216128 metric_logger.py:185] completed step: 18, seconds: 0.117, TFLOP/s/device: 115.982, Tokens/s/device: 17482.010, total_weights: 16384, loss: 1.521 I0416 13:53:43.661673 131281465216128 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0416 13:53:43.662638 131281465216128 checkpointing.py:776] Waited 0.0009987354278564453 seconds for step 19 to finish before starting checkpointing. I0416 13:53:43.662891 131281465216128 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:53:43.966081 131175312000576 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15/items/array_metadatas/process_0 I0416 13:54:08.429058 131173734942272 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 25.248260s (commit=24.815094s, array_metadata_write=0.433166s) I0416 13:54:08.430207 131173787371072 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 378.532 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0416 13:54:08.430264 131173787371072 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.248315s. I0416 13:54:08.814621 131173787371072 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 13:54:09.318025 131173787371072 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.631858s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15/items I0416 13:54:09.318863 131173787371072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15/items I0416 13:54:09.557623 131173787371072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15 I0416 13:54:10.215893 131173787371072 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15`. I0416 13:54:10.216627 131173787371072 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.258006s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/15 I0416 13:54:10.216711 131173787371072 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 27.034763s. I0416 13:54:10.216943 131175259571776 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:54:10.217082 131175259571776 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:54:10.217139 131175259571776 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:54:10.217187 131175259571776 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0416 13:54:10.217382 131281465216128 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0416 13:54:10.217521 131281465216128 checkpoint_manager.py:1441] Waiting for previous save to complete took 26.554633 seconds. If this number is high, consider checkpointing less frequently. I0416 13:54:10.217768 131281465216128 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0416 13:54:10.218095 131281465216128 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19. I0416 13:54:10.406448 131173787371072 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19 I0416 13:54:10.424943 131281465216128 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:54:10.425065 131281465216128 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 13:54:11.095253 131175270057536 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19/items I0416 13:54:20.366475 131281465216128 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.941959s I0416 13:54:20.370978 131281465216128 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.228 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0416 13:54:20.371498 131281465216128 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.048537s (batch_requests_ready=0.098238s, total_serialization_initiated=9.945502s, others=0.004798s) I0416 13:54:20.371593 131281465216128 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.049102s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=10.049083) I0416 13:54:20.373002 131173787371072 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:54:20.373134 131281465216128 async_checkpointer.py:561] Finished blocking save. Time taken: 10.155320s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19. I0416 13:54:20.373354 131281465216128 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:54:20.373614 131175249086016 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:54:20.373780 131281465216128 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347623.6628644, 'wait_for_prev_duration_secs': 26.554632663726807, 'time_between_consecutive_saves_sec': 8.70725965499878, 'checkpointer_blocking_start_time': 1776347650.217792, 'checkpointer_blocking_duration_secs': 10.155477046966553, 'get_old_steps_start_time': 1776347660.373287, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776347623.6628315, 'checkpoint_manager_blocking_duration_secs': 36.71091437339783} I0416 13:54:20.374114 131281465216128 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0416 13:54:20.374156 131281465216128 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:54:21.090775 131175312000576 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19/items/array_metadatas/process_0 I0416 13:54:46.733797 131175270057536 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.362369s (commit=25.916861s, array_metadata_write=0.445508s) I0416 13:54:46.734982 131173787371072 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 347.050 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0416 13:54:46.735059 131173787371072 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.361860s. I0416 13:54:47.140417 131173787371072 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 13:54:47.647368 131173787371072 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.637636s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19/items I0416 13:54:47.648203 131173787371072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19/items I0416 13:54:47.906444 131173787371072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19 I0416 13:54:48.579728 131173787371072 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19`. I0416 13:54:48.580520 131173787371072 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.362715s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_false/checkpoints/19 I0416 13:54:48.580611 131173787371072 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.207424s. I0416 13:54:48.581636 131175249086016 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:54:48.581779 131175249086016 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:54:48.581863 131175249086016 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:54:48.581928 131175249086016 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0416 13:54:48.582088 131281465216128 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0416 13:54:48.582283 131281465216128 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 13:54:48.583222 131281465216128 metric_logger.py:185] completed step: 19, seconds: 0.118, TFLOP/s/device: 115.038, Tokens/s/device: 17339.768, total_weights: 16384, loss: 1.449 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops