MaxView

XPK Start: Thu Apr 23 11:22:57 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
`rope_parameters`'s factor field must be a float >= 1, got 40
`rope_parameters`'s beta_fast field must be a float, got 32
`rope_parameters`'s beta_slow field must be a float, got 1
DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 
2026-04-23 11:23:22.160900: 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 11:23:22.373622 138476343879488 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 11:23:31.415512 138476343879488 distributed.py:149] Starting JAX distributed service on [::]:8482
I0423 11:23:31.417858 138476343879488 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-akd7r-slice-job-0-0.mt-05-fp8-akd7r:8482
I0423 11:23:32.538664 138476343879488 max_utils.py:284] Jax distributed system initialized!
I0423 11:23:38.548303 138476343879488 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 11:23:38.548407 138476343879488 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 11:23:38.548447 138476343879488 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0423 11:23:38.548481 138476343879488 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 11:23:39.242398 138476343879488 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 11:23:39.682112 138476343879488 checkpointing.py:688] Setting up checkpoint logger...
I0423 11:23:39.682232 138476343879488 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0423 11:23:39.682277 138476343879488 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0423 11:23:39.682502 138476343879488 base_pytree_checkpoint_handler.py:441] 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 0x7df0d1948110>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 11:23:42.556371 138476343879488 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0423 11:23:42.556616 138476343879488 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ddc80574170>}, handler_registry=None
I0423 11:23:42.556856 138476343879488 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ddc80574170>` 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 11:23:42.556905 138476343879488 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ddca41233b0>` 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 11:23:42.556941 138476343879488 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 0x7ddc80574170>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ddc80574170>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ddca41233b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ddca41233b0>}).
I0423 11:23:42.557283 138476343879488 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0423 11:23:42.557354 138476343879488 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ddc58702e80> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0423 11:23:43.709947 138476343879488 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints
I0423 11:23:44.026435 138476343879488 checkpoint_manager.py:929] [process=6][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_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, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ddc80517e60>
I0423 11:23:44.026626 138476343879488 checkpointing.py:302] Checkpoint manager created!
I0423 11:23:45.299234 138476343879488 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 11:23:45.299357 138476343879488 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32

I0423 11:23:47.816463 138476343879488 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 11:23:47.816563 138476343879488 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 11:23:47.822216 138476343879488 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 11:23:47.822275 138476343879488 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 11:23:47.838719 138476343879488 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 11:23:47.838777 138476343879488 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 11:23:47.854513 138476343879488 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 11:23:47.854572 138476343879488 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 11:23:47.915698 138476343879488 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 11:23:47.915788 138476343879488 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 11:23:47.931755 138476343879488 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 11:23:47.931815 138476343879488 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 11:23:47.947672 138476343879488 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 11:23:47.947734 138476343879488 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 11:23:47.978347 138476343879488 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 11:23:47.978420 138476343879488 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 11:23:48.039266 138476343879488 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 11:23:48.039353 138476343879488 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0423 11:24:01.902683 138476343879488 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0423 11:24:01.906768 138476343879488 metric_logger.py:301] number parameters: 1.104 billion
I0423 11:24:15.628524 138476343879488 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0423 11:24:15.794578 138476343879488 checkpointing.py:798] Waited 0.1660454273223877 seconds for step 0 to finish before starting checkpointing.
I0423 11:24:15.797031 138476343879488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 11:24:15.799086 138476343879488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0423 11:24:15.800658 138476343879488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/0.
I0423 11:24:16.147910 138476343879488 signaling_client.py:364] Using JaxDistributedSignalingClient
I0423 11:24:16.148951 138476343879488 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0423 11:24:16.149019 138476343879488 replica_slices.py:424] 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 11:24:16.559919 138476343879488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.413723s
I0423 11:24:16.560088 138476343879488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 3.578 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.43112635612487793 s) (per-host)
I0423 11:24:16.560154 138476343879488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.431203s (batch_requests_ready=0.006071s, total_serialization_initiated=0.425048s, others=0.000083s)
I0423 11:24:16.560268 138476343879488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.435347s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.435330)
I0423 11:24:16.561080 138476343879488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/0.
I0423 11:24:16.561447 138350199555840 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 11:44:16.561407
I0423 11:24:16.575229 138476343879488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 11:24:16.575506 138349630580480 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 11:24:16.575628 138476343879488 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776943455.7970119, 'wait_for_prev_duration_secs': 6.794929504394531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776943455.799141, 'checkpointer_blocking_duration_secs': 0.7624106407165527, 'get_old_steps_start_time': 1776943456.5615773, 'get_old_steps_duration_secs': 3.1948089599609375e-05, 'checkpoint_manager_blocking_start_time': 1776943455.795049, 'checkpoint_manager_blocking_duration_secs': 0.7805404663085938}
I0423 11:24:16.575758 138476343879488 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0423 11:24:16.575832 138476343879488 max_utils.py:750] 
Memstats: After params initialized:
I0423 11:24:16.575882 138476343879488 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_24(process=6,(0,6,0,0))
I0423 11:24:16.575913 138476343879488 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_25(process=6,(1,6,0,0))
I0423 11:24:16.575939 138476343879488 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_28(process=6,(0,7,0,0))
I0423 11:24:16.575963 138476343879488 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_29(process=6,(1,7,0,0))
I0423 11:24:16.893136 138476343879488 metric_logger.py:196] completed step: 0, seconds: 13.720, TFLOP/s/device: 0.990, Tokens/s/device: 149.276, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0423 11:24:17.499917 138476343879488 metric_logger.py:196] completed step: 1, seconds: 1.262, TFLOP/s/device: 10.769, Tokens/s/device: 1623.160, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0423 11:24:17.657695 138476343879488 metric_logger.py:196] completed step: 2, seconds: 0.457, TFLOP/s/device: 29.754, Tokens/s/device: 4484.786, total_weights: 65536, loss: 10.856, lm_loss: 10.856, perplexity: 51864.762
I0423 11:24:17.814408 138476343879488 metric_logger.py:196] completed step: 3, seconds: 0.161, TFLOP/s/device: 84.434, Tokens/s/device: 12726.821, total_weights: 65536, loss: 10.824, lm_loss: 10.824, perplexity: 50225.512
I0423 11:24:18.129023 138476343879488 metric_logger.py:196] completed step: 4, seconds: 0.158, TFLOP/s/device: 86.234, Tokens/s/device: 12998.058, total_weights: 65536, loss: 10.793, lm_loss: 10.793, perplexity: 48689.156
I0423 11:24:18.138588 138476343879488 metric_logger.py:196] completed step: 5, seconds: 0.157, TFLOP/s/device: 86.724, Tokens/s/device: 13071.979, total_weights: 65536, loss: 10.763, lm_loss: 10.763, perplexity: 47229.012
I0423 11:24:19.353344    2797 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0423 11:24:21.397144 138349638973184 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/0/items/array_metadatas/process_6
I0423 11:24:34.620592 138350199555840 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 85.431 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 18.491591930389404 s) (per-host)
I0423 11:24:34.620720 138350199555840 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 18.058602s.
I0423 11:24:42.226716 138476343879488 metric_logger.py:196] completed step: 6, seconds: 0.315, TFLOP/s/device: 43.147, Tokens/s/device: 6503.528, total_weights: 65536, loss: 10.733, lm_loss: 10.733, perplexity: 45864.258
I0423 11:24:42.383740 138476343879488 metric_logger.py:196] completed step: 7, seconds: 23.935, TFLOP/s/device: 0.568, Tokens/s/device: 85.566, total_weights: 65536, loss: 10.706, lm_loss: 10.706, perplexity: 44620.242
I0423 11:24:42.540900 138476343879488 metric_logger.py:196] completed step: 8, seconds: 0.163, TFLOP/s/device: 83.389, Tokens/s/device: 12569.275, total_weights: 65536, loss: 10.680, lm_loss: 10.680, perplexity: 43477.055
I0423 11:24:42.549367 138476343879488 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0423 11:24:42.854953 138476343879488 checkpointing.py:798] Waited 0.3055562973022461 seconds for step 10 to finish before starting checkpointing.
I0423 11:24:42.857391 138476343879488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 11:24:44.951512 138350199555840 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 28.389377s.
I0423 11:24:44.951737 138349630580480 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 11:24:44.951794 138349630580480 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 11:24:44.951844 138349630580480 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0423 11:24:44.953970 138349630580480 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0423 11:24:44.954136 138476343879488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0423 11:24:44.954277 138476343879488 checkpoint_manager.py:1452] Waiting for previous save to complete took 2.096879 seconds. If this number is high, consider checkpointing less frequently.
I0423 11:24:44.956278 138476343879488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0423 11:24:44.958257 138476343879488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/10.
I0423 11:24:45.261173 138476343879488 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0423 11:24:45.261279 138476343879488 replica_slices.py:424] 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 11:24:45.313155 138476343879488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054641s
I0423 11:24:45.313330 138476343879488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 22.307 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06916046142578125 s) (per-host)
I0423 11:24:45.313380 138476343879488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.069221s (batch_requests_ready=0.005174s, total_serialization_initiated=0.063979s, others=0.000068s)
I0423 11:24:45.313500 138476343879488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.073557s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.073541)
I0423 11:24:45.314230 138476343879488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/10.
I0423 11:24:45.314535 138349630580480 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 11:44:45.314503
I0423 11:24:45.750123 138476343879488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 11:24:45.750495 138347515201280 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 11:24:45.750667 138476343879488 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776943482.8573606, 'wait_for_prev_duration_secs': 2.0968785285949707, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776943484.9563165, 'checkpointer_blocking_duration_secs': 0.3583698272705078, 'get_old_steps_start_time': 1776943485.3147094, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776943482.8553982, 'checkpoint_manager_blocking_duration_secs': 2.8952322006225586}
I0423 11:24:45.750786 138476343879488 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0423 11:24:45.751546 138476343879488 metric_logger.py:196] completed step: 9, seconds: 0.157, TFLOP/s/device: 86.550, Tokens/s/device: 13045.749, total_weights: 65536, loss: 10.656, lm_loss: 10.656, perplexity: 42467.320
I0423 11:24:45.763238 138476343879488 metric_logger.py:196] completed step: 10, seconds: 0.157, TFLOP/s/device: 86.460, Tokens/s/device: 13032.135, total_weights: 65536, loss: 10.636, lm_loss: 10.636, perplexity: 41601.941
I0423 11:24:45.915254 138476343879488 metric_logger.py:196] completed step: 11, seconds: 3.212, TFLOP/s/device: 4.231, Tokens/s/device: 637.681, total_weights: 65536, loss: 10.618, lm_loss: 10.618, perplexity: 40849.570
I0423 11:24:46.485357 138476343879488 metric_logger.py:196] completed step: 12, seconds: 0.009, TFLOP/s/device: 1467.293, Tokens/s/device: 221166.307, total_weights: 65536, loss: 10.602, lm_loss: 10.602, perplexity: 40203.926
I0423 11:24:46.642166 138476343879488 metric_logger.py:196] completed step: 13, seconds: 0.560, TFLOP/s/device: 24.248, Tokens/s/device: 3654.956, total_weights: 65536, loss: 10.588, lm_loss: 10.588, perplexity: 39652.145
I0423 11:24:46.799124 138476343879488 metric_logger.py:196] completed step: 14, seconds: 0.163, TFLOP/s/device: 83.173, Tokens/s/device: 12536.729, total_weights: 65536, loss: 10.577, lm_loss: 10.577, perplexity: 39212.879
I0423 11:24:46.955974 138476343879488 metric_logger.py:196] completed step: 15, seconds: 0.157, TFLOP/s/device: 86.738, Tokens/s/device: 13074.149, total_weights: 65536, loss: 10.568, lm_loss: 10.568, perplexity: 38879.223
I0423 11:24:47.112968 138476343879488 metric_logger.py:196] completed step: 16, seconds: 0.157, TFLOP/s/device: 86.545, Tokens/s/device: 13045.001, total_weights: 65536, loss: 10.562, lm_loss: 10.562, perplexity: 38619.105
I0423 11:24:47.269886 138476343879488 metric_logger.py:196] completed step: 17, seconds: 0.157, TFLOP/s/device: 86.729, Tokens/s/device: 13072.730, total_weights: 65536, loss: 10.556, lm_loss: 10.556, perplexity: 38395.816
I0423 11:24:47.426603 138476343879488 metric_logger.py:196] completed step: 18, seconds: 0.157, TFLOP/s/device: 86.472, Tokens/s/device: 13034.043, total_weights: 65536, loss: 10.552, lm_loss: 10.552, perplexity: 38236.652
I0423 11:24:47.585217 138476343879488 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0423 11:24:47.586515 138476343879488 checkpointing.py:798] Waited 0.0013074874877929688 seconds for step 19 to finish before starting checkpointing.
I0423 11:24:47.588700 138476343879488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 11:24:51.144377 138346466637568 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/10/items/array_metadatas/process_6
I0423 11:25:07.448043 138349630580480 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 71.148 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 22.203851461410522 s) (per-host)
I0423 11:25:07.448170 138349630580480 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 22.133517s.
I0423 11:25:14.874703 138349630580480 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 29.560034s.
I0423 11:25:14.874973 138347515201280 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 11:25:14.875108 138347515201280 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 11:25:14.875159 138347515201280 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0423 11:25:14.876787 138347515201280 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0423 11:25:14.876959 138476343879488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0423 11:25:14.877085 138476343879488 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.288389 seconds. If this number is high, consider checkpointing less frequently.
I0423 11:25:14.878758 138476343879488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0423 11:25:14.880938 138476343879488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/19.
I0423 11:25:16.061814 138476343879488 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0423 11:25:16.061923 138476343879488 replica_slices.py:424] 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 11:25:16.113752 138476343879488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054881s
I0423 11:25:16.113916 138476343879488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 2.955 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5221428871154785 s) (per-host)
I0423 11:25:16.113964 138476343879488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.522201s (batch_requests_ready=0.005087s, total_serialization_initiated=0.517051s, others=0.000063s)
I0423 11:25:16.114076 138476343879488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.526410s (all_items=0.000011s, per_item={'items': '0.00001097'}, temp_paths=0.526399)
I0423 11:25:16.114814 138476343879488 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.24 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/19.
I0423 11:25:16.115173 138347515201280 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 11:45:16.115136
I0423 11:25:16.117287 138476343879488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 11:25:16.117523 138349630580480 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 11:25:16.117684 138476343879488 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776943487.5886712, 'wait_for_prev_duration_secs': 27.288389444351196, 'time_between_consecutive_saves_sec': 2.634662389755249, 'checkpointer_blocking_start_time': 1776943514.878797, 'checkpointer_blocking_duration_secs': 1.2364819049835205, 'get_old_steps_start_time': 1776943516.1152976, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776943487.5868661, 'checkpoint_manager_blocking_duration_secs': 28.53078317642212}
I0423 11:25:16.117801 138476343879488 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0423 11:25:16.117842 138476343879488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 11:25:20.735120 138347531986688 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260423_093806_05_fp8/checkpoints/19/items/array_metadatas/process_6
I0423 11:25:36.698059 138347515201280 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 74.848 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.106242895126343 s) (per-host)
I0423 11:25:36.698184 138347515201280 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.582349s.
I0423 11:25:43.985370 138347515201280 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 27.869524s.
I0423 11:25:43.985633 138349630580480 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 11:25:43.985765 138349630580480 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 11:25:43.985809 138349630580480 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0423 11:25:43.987157 138349630580480 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0423 11:25:43.987318 138476343879488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0423 11:25:43.987501 138476343879488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 11:25:43.988426 138476343879488 metric_logger.py:196] completed step: 19, seconds: 0.157, TFLOP/s/device: 86.543, Tokens/s/device: 13044.669, total_weights: 65536, loss: 10.549, lm_loss: 10.549, perplexity: 38125.074
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Thu Apr 23 11:25:52 UTC 2026
EXIT_CODE=0