XPK Start: Sun Apr 19 01:26:09 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 01:26:33.228542: 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 01:26:33.404891 133027606705984 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 01:26:42.445465 133027606705984 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 01:26:42.447852 133027606705984 distributed.py:172] Connecting to JAX distributed service on mt-02-synthetic-qzqg6-slice-job-0-0.mt-02-synthetic-qzqg6:8482 I0419 01:26:43.947919 133027606705984 max_utils.py:284] Jax distributed system initialized! I0419 01:26:50.187024 133027606705984 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 01:26:50.187125 133027606705984 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 01:26:50.187164 133027606705984 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 01:26:50.187197 133027606705984 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 01:26:51.241559 133027606705984 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 01:26:51.353365 133027606705984 checkpointing.py:688] Setting up checkpoint logger... I0419 01:26:51.353479 133027606705984 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 01:26:51.353525 133027606705984 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 01:26:51.353722 133027606705984 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 0x78fc443182c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 01:26:55.881212 133027606705984 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 01:26:55.881458 133027606705984 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 0x78fb24cfccb0>}, handler_registry=None I0419 01:26:55.881690 133027606705984 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78fb24cfccb0>` 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 01:26:55.881740 133027606705984 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78fb23c5da30>` 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 01:26:55.881777 133027606705984 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 0x78fb24cfccb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78fb24cfccb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78fb23c5da30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78fb23c5da30>}). I0419 01:26:55.882094 133027606705984 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 01:26:55.882164 133027606705984 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78e7202bc4a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 01:26:56.972307 133027606705984 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_02_synthetic/checkpoints I0419 01:26:57.444163 133027606705984 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_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78fb23d9a450> I0419 01:26:57.444362 133027606705984 checkpointing.py:302] Checkpoint manager created! I0419 01:26:58.430489 133027606705984 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 01:26:58.430611 133027606705984 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 01:26:59.827447 133027606705984 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 01:26:59.827587 133027606705984 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:26:59.827651 133027606705984 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:26:59.827697 133027606705984 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 01:26:59.827733 133027606705984 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:26:59.827768 133027606705984 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:26:59.827803 133027606705984 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:26:59.827836 133027606705984 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 01:26:59.827889 133027606705984 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:26:59.827942 133027606705984 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:26:59.828001 133027606705984 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 01:26:59.828052 133027606705984 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 01:27:00.226018 133027606705984 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:27:00.226107 133027606705984 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:27:00.232060 133027606705984 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:27:00.232126 133027606705984 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:27:00.248692 133027606705984 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:27:00.248749 133027606705984 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:27:00.264568 133027606705984 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:27:00.264626 133027606705984 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:27:00.288924 133027606705984 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 01:27:00.288990 133027606705984 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:27:00.304989 133027606705984 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 01:27:00.305051 133027606705984 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:27:00.320958 133027606705984 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 01:27:00.321018 133027606705984 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:27:00.350995 133027606705984 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 01:27:00.351065 133027606705984 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:27:00.371043 133027606705984 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 01:27:00.371106 133027606705984 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 01:27:11.773020 133027606705984 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. I0419 01:27:11.775411 133027606705984 metric_logger.py:301] number parameters: 1.104 billion I0419 01:27:23.039474 133027606705984 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 01:27:23.190600 133027606705984 checkpointing.py:798] Waited 0.15110993385314941 seconds for step 0 to finish before starting checkpointing. I0419 01:27:23.192898 133027606705984 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:27:23.194599 133027606705984 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0419 01:27:23.196084 133027606705984 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_02_synthetic/checkpoints/0. I0419 01:27:23.972692 133027606705984 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 01:27:23.973737 133027606705984 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 01:27:23.973795 133027606705984 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 01:27:24.254070 133027606705984 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282152s I0419 01:27:24.254266 133027606705984 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.299 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29110240936279297 s) (per-host) I0419 01:27:24.254322 133027606705984 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.291190s (batch_requests_ready=0.002959s, total_serialization_initiated=0.288136s, others=0.000094s) I0419 01:27:24.254424 133027606705984 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.295633s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.295616) I0419 01:27:24.255166 133027606705984 event_tracking.py:125] [process=3] [async] Finished blocking save in 1.06 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_02_synthetic/checkpoints/0. I0419 01:27:24.255520 132898817652480 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:47:24.255481 I0419 01:27:24.258918 133027606705984 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:27:24.259202 132895452686080 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:27:24.259377 133027606705984 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_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776562043.192868, 'wait_for_prev_duration_secs': 9.083747863769531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776562043.1946564, 'checkpointer_blocking_duration_secs': 1.0610113143920898, 'get_old_steps_start_time': 1776562044.2556918, 'get_old_steps_duration_secs': 3.0994415283203125e-05, 'checkpoint_manager_blocking_start_time': 1776562043.1910064, 'checkpoint_manager_blocking_duration_secs': 1.0683350563049316} I0419 01:27:24.259482 133027606705984 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 01:27:24.259549 133027606705984 max_utils.py:750] Memstats: After params initialized: I0419 01:27:24.259597 133027606705984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0419 01:27:24.259628 133027606705984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0419 01:27:24.259655 133027606705984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0419 01:27:24.259680 133027606705984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0419 01:27:24.580114 133027606705984 metric_logger.py:196] completed step: 0, seconds: 11.263, TFLOP/s/device: 1.206, Tokens/s/device: 181.835, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 01:27:24.753908 133027606705984 metric_logger.py:196] completed step: 1, seconds: 1.539, TFLOP/s/device: 8.830, Tokens/s/device: 1330.968, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 01:27:25.176887 133027606705984 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 496.933, Tokens/s/device: 74903.080, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28632.133 I0419 01:27:25.325378 133027606705984 metric_logger.py:196] completed step: 3, seconds: 0.413, TFLOP/s/device: 32.918, Tokens/s/device: 4961.721, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16855.789 I0419 01:27:25.621422 133027606705984 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.365, Tokens/s/device: 12414.980, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10661.594 I0419 01:27:25.628900 133027606705984 metric_logger.py:196] completed step: 5, seconds: 0.148, TFLOP/s/device: 91.572, Tokens/s/device: 13802.678, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7245.978 I0419 01:27:30.114088 2819 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 01:27:34.771382 132895561266944 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 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_02_synthetic/checkpoints/0/items/array_metadatas/process_3 I0419 01:27:49.947287 133027606705984 metric_logger.py:196] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.884, Tokens/s/device: 6916.209, total_weights: 65536, loss: 8.589, lm_loss: 8.589, perplexity: 5371.156 I0419 01:27:50.095138 133027606705984 metric_logger.py:196] completed step: 7, seconds: 24.172, TFLOP/s/device: 0.562, Tokens/s/device: 84.727, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4360.465 I0419 01:27:50.242839 133027606705984 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.701, Tokens/s/device: 13369.979, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.256 I0419 01:27:50.391411 133027606705984 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 01:27:50.392058 133027606705984 checkpointing.py:798] Waited 0.000659942626953125 seconds for step 9 to finish before starting checkpointing. I0419 01:27:50.394185 133027606705984 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:28:03.752175 132898817652480 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 39.699 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 39.78899621963501 s) (per-host) I0419 01:28:03.752318 132898817652480 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.496688s. I0419 01:28:15.970752 132898817652480 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 51.715106s. I0419 01:28:15.971018 132895452686080 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:28:15.971147 132895452686080 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:28:15.971199 132895452686080 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:28:15.974050 132895452686080 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 01:28:15.974248 133027606705984 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 01:28:15.974380 133027606705984 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.580199 seconds. If this number is high, consider checkpointing less frequently. I0419 01:28:15.976611 133027606705984 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0419 01:28:15.978565 133027606705984 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_02_synthetic/checkpoints/9. I0419 01:28:17.127249 133027606705984 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 01:28:17.127343 133027606705984 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 01:28:17.160409 133027606705984 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034798s I0419 01:28:17.160583 133027606705984 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 36.928 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04177212715148926 s) (per-host) I0419 01:28:17.160634 133027606705984 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.041833s (batch_requests_ready=0.002655s, total_serialization_initiated=0.039109s, others=0.000068s) I0419 01:28:17.160742 133027606705984 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.046163s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.046147) I0419 01:28:17.161494 133027606705984 event_tracking.py:125] [process=3] [async] Finished blocking save in 1.18 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_02_synthetic/checkpoints/9. I0419 01:28:17.161791 132895452686080 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:48:17.161757 I0419 01:28:17.164195 133027606705984 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:28:17.164434 132895561266944 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:28:17.164556 133027606705984 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_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776562070.3941553, 'wait_for_prev_duration_secs': 25.580198764801025, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776562095.9766493, 'checkpointer_blocking_duration_secs': 1.185291051864624, 'get_old_steps_start_time': 1776562097.161963, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776562070.3923008, 'checkpoint_manager_blocking_duration_secs': 26.772221088409424} I0419 01:28:17.164672 133027606705984 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 01:28:17.164712 133027606705984 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:28:25.616336 132876976756480 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 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_02_synthetic/checkpoints/9/items/array_metadatas/process_3 I0419 01:29:01.977301 132895452686080 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 35.213 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.85843777656555 s) (per-host) I0419 01:29:01.977429 132895452686080 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.815520s. I0419 01:29:12.344323 132895452686080 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 55.182399s. I0419 01:29:12.344531 132895561266944 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:29:12.344588 132895561266944 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:29:12.344633 132895561266944 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:29:12.346486 132895561266944 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 01:29:12.346660 133027606705984 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 01:29:12.346832 133027606705984 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:29:12.347747 133027606705984 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.612, Tokens/s/device: 13808.728, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3551.438 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 01:29:25 UTC 2026 EXIT_CODE=0