XPK Start: Sun Apr 19 01:30:59 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 01:31:23.239994: 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) I0419 01:31:23.418859 133315306633024 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 01:31:32.458225 133315306633024 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 01:31:32.460580 133315306633024 distributed.py:172] Connecting to JAX distributed service on mt-04-int8-ztoii-slice-job-0-0.mt-04-int8-ztoii:8482 I0419 01:31:33.904474 133315306633024 max_utils.py:284] Jax distributed system initialized! I0419 01:31:39.977716 133315306633024 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 01:31:39.977820 133315306633024 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 01:31:39.977860 133315306633024 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 I0419 01:31:39.977890 133315306633024 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 01:31:41.040132 133315306633024 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 01:31:41.588927 133315306633024 checkpointing.py:688] Setting up checkpoint logger... I0419 01:31:41.589042 133315306633024 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 01:31:41.589087 133315306633024 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 01:31:41.589319 133315306633024 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 0x793f2bc4d0a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 01:31:45.227886 133315306633024 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 01:31:45.228153 133315306633024 checkpoint_manager.py:708] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x793e1ffae270>}, handler_registry=None I0419 01:31:45.228431 133315306633024 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x793e1ffae270>` 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`. I0419 01:31:45.228493 133315306633024 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792b144ab4a0>` 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`. I0419 01:31:45.228549 133315306633024 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 0x793e1ffae270>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x793e1ffae270>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792b144ab4a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792b144ab4a0>}). I0419 01:31:45.228893 133315306633024 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 01:31:45.228984 133315306633024 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x792b143f1d00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 01:31:46.915183 133315306633024 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints I0419 01:31:47.367382 133315306633024 checkpoint_manager.py:929] [process=3][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_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x792afc73ff50> I0419 01:31:47.367556 133315306633024 checkpointing.py:302] Checkpoint manager created! I0419 01:31:48.735786 133315306633024 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 01:31:48.735914 133315306633024 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 01:31:51.038955 133315306633024 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 01:31:51.039072 133315306633024 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:31:51.039124 133315306633024 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:31:51.039167 133315306633024 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 01:31:51.039205 133315306633024 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:31:51.039252 133315306633024 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:31:51.039287 133315306633024 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:31:51.039322 133315306633024 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 01:31:51.039364 133315306633024 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:31:51.039399 133315306633024 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:31:51.039430 133315306633024 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 01:31:51.039462 133315306633024 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 01:31:51.455967 133315306633024 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:31:51.456059 133315306633024 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:31:51.461200 133315306633024 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:31:51.461280 133315306633024 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:31:51.477676 133315306633024 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:31:51.477737 133315306633024 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:31:51.493445 133315306633024 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:31:51.493502 133315306633024 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:31:51.574490 133315306633024 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:31:51.574589 133315306633024 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:31:51.590563 133315306633024 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:31:51.590624 133315306633024 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:31:51.606445 133315306633024 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:31:51.606507 133315306633024 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:31:51.636559 133315306633024 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 01:31:51.636631 133315306633024 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:31:51.715104 133315306633024 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 01:31:51.715193 133315306633024 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 01:32:06.230765 133315306633024 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0419 01:32:06.234261 133315306633024 metric_logger.py:301] number parameters: 1.104 billion I0419 01:32:20.823879 133315306633024 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 01:32:21.063760 133315306633024 checkpointing.py:798] Waited 0.2398688793182373 seconds for step 0 to finish before starting checkpointing. I0419 01:32:21.066429 133315306633024 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:32:21.068362 133315306633024 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0419 01:32:21.069796 133315306633024 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/0. I0419 01:32:21.801306 133315306633024 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 01:32:21.802308 133315306633024 jax_array_handlers.py:360] Scheduling D2H of 111 prioritized jax.Array. I0419 01:32:21.802368 133315306633024 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 I0419 01:32:22.084192 133315306633024 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.284497s I0419 01:32:22.084383 133315306633024 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.158 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2990889549255371 s) (per-host) I0419 01:32:22.084434 133315306633024 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.299150s (batch_requests_ready=0.004472s, total_serialization_initiated=0.294609s, others=0.000068s) I0419 01:32:22.084549 133315306633024 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.303650s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.303633) I0419 01:32:22.085353 133315306633024 event_tracking.py:125] [process=3] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/0. I0419 01:32:22.085679 133180581463808 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:52:22.085639 I0419 01:32:22.098030 133315306633024 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:32:22.098304 133183837390592 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:32:22.098459 133315306633024 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776562341.066411, 'wait_for_prev_duration_secs': 6.151199340820312e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776562341.068401, 'checkpointer_blocking_duration_secs': 1.017437219619751, 'get_old_steps_start_time': 1776562342.0858636, 'get_old_steps_duration_secs': 3.1948089599609375e-05, 'checkpoint_manager_blocking_start_time': 1776562341.0644228, 'checkpoint_manager_blocking_duration_secs': 1.0339975357055664} I0419 01:32:22.098579 133315306633024 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 01:32:22.098655 133315306633024 max_utils.py:750] Memstats: After params initialized: I0419 01:32:22.098704 133315306633024 max_utils.py:756] Using (GB) 0.44 / 31.25 (1.408000%) on TPU_10(process=3,(2,2,0,0)) I0419 01:32:22.098737 133315306633024 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0419 01:32:22.098764 133315306633024 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0419 01:32:22.098788 133315306633024 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0419 01:32:22.431117 133315306633024 metric_logger.py:196] completed step: 0, seconds: 14.588, TFLOP/s/device: 0.931, Tokens/s/device: 140.392, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52668.902 I0419 01:32:22.682760 133315306633024 metric_logger.py:196] completed step: 1, seconds: 1.605, TFLOP/s/device: 8.466, Tokens/s/device: 1276.148, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52668.902 I0419 01:32:23.125355 133315306633024 metric_logger.py:196] completed step: 2, seconds: 0.026, TFLOP/s/device: 522.341, Tokens/s/device: 78732.893, total_weights: 65536, loss: 10.260, lm_loss: 10.260, perplexity: 28557.658 I0419 01:32:23.354916 133315306633024 metric_logger.py:196] completed step: 3, seconds: 0.421, TFLOP/s/device: 32.244, Tokens/s/device: 4860.164, total_weights: 65536, loss: 9.728, lm_loss: 9.728, perplexity: 16789.145 I0419 01:32:23.843251 133315306633024 metric_logger.py:196] completed step: 4, seconds: 0.256, TFLOP/s/device: 52.988, Tokens/s/device: 7986.959, total_weights: 65536, loss: 9.271, lm_loss: 9.271, perplexity: 10625.563 I0419 01:32:23.852307 133315306633024 metric_logger.py:196] completed step: 5, seconds: 0.229, TFLOP/s/device: 59.209, Tokens/s/device: 8924.642, total_weights: 65536, loss: 8.885, lm_loss: 8.885, perplexity: 7225.917 I0419 01:32:26.566681 2961 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 01:32:30.107187 133180589856512 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 111 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/0/items/array_metadatas/process_3 I0419 01:32:53.362218 133315306633024 metric_logger.py:196] completed step: 6, seconds: 0.489, TFLOP/s/device: 27.801, Tokens/s/device: 4190.521, total_weights: 65536, loss: 8.586, lm_loss: 8.586, perplexity: 5354.806 I0419 01:32:53.591745 133315306633024 metric_logger.py:196] completed step: 7, seconds: 29.283, TFLOP/s/device: 0.464, Tokens/s/device: 69.938, total_weights: 65536, loss: 8.378, lm_loss: 8.378, perplexity: 4350.756 I0419 01:32:53.821067 133315306633024 metric_logger.py:196] completed step: 8, seconds: 0.236, TFLOP/s/device: 57.659, Tokens/s/device: 8690.929, total_weights: 65536, loss: 8.249, lm_loss: 8.249, perplexity: 3824.619 I0419 01:32:54.052030 133315306633024 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 01:32:54.052967 133315306633024 checkpointing.py:798] Waited 0.000949859619140625 seconds for step 9 to finish before starting checkpointing. I0419 01:32:54.055485 133315306633024 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:32:56.640728 133180581463808 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 45.319 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 34.85538196563721 s) (per-host) I0419 01:32:56.640855 133180581463808 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.555062s. I0419 01:33:06.836740 133180581463808 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 44.750932s. I0419 01:33:06.837014 133183837390592 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:33:06.837157 133183837390592 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:33:06.837222 133183837390592 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:33:06.839336 133183837390592 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 01:33:06.839500 133315306633024 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 01:33:06.839637 133315306633024 checkpoint_manager.py:1452] Waiting for previous save to complete took 12.784148 seconds. If this number is high, consider checkpointing less frequently. I0419 01:33:06.841774 133315306633024 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0419 01:33:06.843775 133315306633024 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/9. I0419 01:33:07.570105 133315306633024 jax_array_handlers.py:360] Scheduling D2H of 111 prioritized jax.Array. I0419 01:33:07.570202 133315306633024 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 I0419 01:33:07.616373 133315306633024 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.048643s I0419 01:33:07.616541 133315306633024 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 25.466 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06057286262512207 s) (per-host) I0419 01:33:07.616590 133315306633024 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.060632s (batch_requests_ready=0.004240s, total_serialization_initiated=0.056325s, others=0.000067s) I0419 01:33:07.616706 133315306633024 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.064833s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.064818) I0419 01:33:07.617473 133315306633024 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/9. I0419 01:33:07.617775 133183837390592 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:53:07.617742 I0419 01:33:07.619775 133315306633024 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:33:07.620075 133180589856512 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:33:07.620213 133315306633024 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776562374.055456, 'wait_for_prev_duration_secs': 12.784148216247559, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776562386.841813, 'checkpointer_blocking_duration_secs': 0.776111364364624, 'get_old_steps_start_time': 1776562387.6179495, 'get_old_steps_duration_secs': 3.24249267578125e-05, 'checkpoint_manager_blocking_start_time': 1776562374.0532575, 'checkpoint_manager_blocking_duration_secs': 13.566921949386597} I0419 01:33:07.620405 133315306633024 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 01:33:07.620452 133315306633024 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:33:15.374760 133160419452672 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 111 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_04_int8/checkpoints/9/items/array_metadatas/process_3 I0419 01:33:50.432538 133183837390592 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 36.841 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.87655067443848 s) (per-host) I0419 01:33:50.432642 133183837390592 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.814752s. I0419 01:33:59.039895 133183837390592 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 51.421990s. I0419 01:33:59.040179 133180589856512 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:33:59.040354 133180589856512 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:33:59.040406 133180589856512 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:33:59.042219 133180589856512 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 01:33:59.042423 133315306633024 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 01:33:59.042612 133315306633024 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:33:59.043525 133315306633024 metric_logger.py:196] completed step: 9, seconds: 0.229, TFLOP/s/device: 59.269, Tokens/s/device: 8933.635, total_weights: 65536, loss: 8.174, lm_loss: 8.174, perplexity: 3546.104 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 01:34:10 UTC 2026 EXIT_CODE=0