XPK Start: Sat Apr 18 22:25:43 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-18 22:26:07.184870: 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) I0418 22:26:07.364533 134083346794304 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 22:26:16.404522 134083346794304 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 22:26:16.406857 134083346794304 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-amtb5-slice-job-0-0.mt-05-fp8-amtb5:8482 I0418 22:26:18.584498 134083346794304 max_utils.py:284] Jax distributed system initialized! I0418 22:26:23.619636 134083346794304 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 22:26:23.619736 134083346794304 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 22:26:23.619776 134083346794304 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 I0418 22:26:23.619812 134083346794304 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 22:26:24.687263 134083346794304 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 22:26:25.115824 134083346794304 checkpointing.py:688] Setting up checkpoint logger... I0418 22:26:25.115952 134083346794304 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 22:26:25.116009 134083346794304 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 22:26:25.116264 134083346794304 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 0x79f21a51c230>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 22:26:29.012442 134083346794304 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0418 22:26:29.012660 134083346794304 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 0x79dea01e9910>}, handler_registry=None I0418 22:26:29.012921 134083346794304 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79dea01e9910>` 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`. I0418 22:26:29.012981 134083346794304 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79de684f9520>` 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`. I0418 22:26:29.013033 134083346794304 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 0x79dea01e9910>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79dea01e9910>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79de684f9520>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79de684f9520>}). I0418 22:26:29.013399 134083346794304 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 22:26:29.013489 134083346794304 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x79de8872b4c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 22:26:29.701782 134083346794304 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints I0418 22:26:30.575011 134083346794304 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_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79de887e8830> I0418 22:26:30.575187 134083346794304 checkpointing.py:302] Checkpoint manager created! I0418 22:26:31.845160 134083346794304 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0418 22:26:31.845294 134083346794304 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 22:26:33.948520 134083346794304 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0418 22:26:33.948637 134083346794304 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0418 22:26:33.948688 134083346794304 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0418 22:26:33.948731 134083346794304 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0418 22:26:33.948767 134083346794304 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0418 22:26:33.948801 134083346794304 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0418 22:26:33.948837 134083346794304 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0418 22:26:33.948871 134083346794304 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0418 22:26:33.948905 134083346794304 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0418 22:26:33.948938 134083346794304 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0418 22:26:33.948971 134083346794304 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0418 22:26:33.949001 134083346794304 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0418 22:26:34.367632 134083346794304 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 22:26:34.367722 134083346794304 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 22:26:34.373284 134083346794304 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 22:26:34.373337 134083346794304 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 22:26:34.389621 134083346794304 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 22:26:34.389680 134083346794304 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 22:26:34.405287 134083346794304 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 22:26:34.405342 134083346794304 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 22:26:34.465301 134083346794304 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 22:26:34.465384 134083346794304 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 22:26:34.481125 134083346794304 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 22:26:34.481185 134083346794304 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 22:26:34.496764 134083346794304 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 22:26:34.496823 134083346794304 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 22:26:34.526847 134083346794304 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 22:26:34.526920 134083346794304 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 22:26:34.586670 134083346794304 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 22:26:34.586748 134083346794304 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 22:26:48.600697 134083346794304 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. I0418 22:26:48.604555 134083346794304 metric_logger.py:301] number parameters: 1.104 billion I0418 22:27:02.317248 134083346794304 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0418 22:27:02.479829 134083346794304 checkpointing.py:798] Waited 0.16257739067077637 seconds for step 0 to finish before starting checkpointing. I0418 22:27:02.482132 134083346794304 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 22:27:02.484210 134083346794304 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0418 22:27:02.485535 134083346794304 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0. I0418 22:27:03.684999 134083346794304 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 22:27:03.685844 134083346794304 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0418 22:27:03.685911 134083346794304 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 I0418 22:27:04.088876 134083346794304 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.405507s I0418 22:27:04.089042 134083346794304 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 3.648 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.4228549003601074 s) (per-host) I0418 22:27:04.089095 134083346794304 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.422918s (batch_requests_ready=0.005599s, total_serialization_initiated=0.417248s, others=0.000070s) I0418 22:27:04.089213 134083346794304 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.427044s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.427026) I0418 22:27:04.090062 134083346794304 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0. I0418 22:27:04.090352 133957132056320 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 22:47:04.090321 I0418 22:27:04.106623 134083346794304 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 22:27:04.106929 133957102315264 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0418 22:27:04.107086 134083346794304 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776551222.4821136, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776551222.4842646, 'checkpointer_blocking_duration_secs': 1.606231451034546, 'get_old_steps_start_time': 1776551224.0905216, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776551222.4803245, 'checkpoint_manager_blocking_duration_secs': 1.6267218589782715} I0418 22:27:04.107205 134083346794304 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0418 22:27:04.107299 134083346794304 max_utils.py:750] Memstats: After params initialized: I0418 22:27:04.107350 134083346794304 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_24(process=6,(0,6,0,0)) I0418 22:27:04.107382 134083346794304 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_25(process=6,(1,6,0,0)) I0418 22:27:04.107409 134083346794304 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_28(process=6,(0,7,0,0)) I0418 22:27:04.107434 134083346794304 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_29(process=6,(1,7,0,0)) I0418 22:27:04.422835 134083346794304 metric_logger.py:196] completed step: 0, seconds: 13.711, TFLOP/s/device: 0.991, Tokens/s/device: 149.374, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742 I0418 22:27:04.616391 134083346794304 metric_logger.py:196] completed step: 1, seconds: 2.103, TFLOP/s/device: 6.461, Tokens/s/device: 973.913, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742 I0418 22:27:05.058399 134083346794304 metric_logger.py:196] completed step: 2, seconds: 0.042, TFLOP/s/device: 323.619, Tokens/s/device: 48779.326, total_weights: 65536, loss: 10.856, lm_loss: 10.856, perplexity: 51864.762 I0418 22:27:05.216041 134083346794304 metric_logger.py:196] completed step: 3, seconds: 0.417, TFLOP/s/device: 32.549, Tokens/s/device: 4906.071, total_weights: 65536, loss: 10.824, lm_loss: 10.824, perplexity: 50225.512 I0418 22:27:05.530991 134083346794304 metric_logger.py:196] completed step: 4, seconds: 0.187, TFLOP/s/device: 72.668, Tokens/s/device: 10953.277, total_weights: 65536, loss: 10.793, lm_loss: 10.793, perplexity: 48689.156 I0418 22:27:05.540683 134083346794304 metric_logger.py:196] completed step: 5, seconds: 0.157, TFLOP/s/device: 86.327, Tokens/s/device: 13012.180, total_weights: 65536, loss: 10.763, lm_loss: 10.763, perplexity: 47229.012 I0418 22:27:08.158208 2798 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 22:27:10.160298 133957652743936 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_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0/items/array_metadatas/process_6 I0418 22:27:24.997488 133957132056320 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 74.058 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.331260681152344 s) (per-host) I0418 22:27:24.997607 133957132056320 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.907146s. I0418 22:27:29.463299 134083346794304 metric_logger.py:196] completed step: 6, seconds: 0.315, TFLOP/s/device: 43.101, Tokens/s/device: 6496.679, total_weights: 65536, loss: 10.733, lm_loss: 10.733, perplexity: 45864.258 I0418 22:27:29.620326 134083346794304 metric_logger.py:196] completed step: 7, seconds: 23.769, TFLOP/s/device: 0.572, Tokens/s/device: 86.161, total_weights: 65536, loss: 10.706, lm_loss: 10.706, perplexity: 44620.242 I0418 22:27:29.777451 134083346794304 metric_logger.py:196] completed step: 8, seconds: 0.163, TFLOP/s/device: 83.438, Tokens/s/device: 12576.685, total_weights: 65536, loss: 10.680, lm_loss: 10.680, perplexity: 43477.055 I0418 22:27:29.785753 134083346794304 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0418 22:27:30.091767 134083346794304 checkpointing.py:798] Waited 0.3059847354888916 seconds for step 10 to finish before starting checkpointing. I0418 22:27:30.094126 134083346794304 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 22:27:34.062683 133957132056320 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 29.972206s. I0418 22:27:34.062945 133957102315264 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 22:27:34.063074 133957102315264 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0418 22:27:34.063125 133957102315264 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 22:27:34.064910 133957102315264 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 22:27:34.065140 134083346794304 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. W0418 22:27:34.065290 134083346794304 checkpoint_manager.py:1452] Waiting for previous save to complete took 3.971163 seconds. If this number is high, consider checkpointing less frequently. I0418 22:27:34.066808 134083346794304 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0418 22:27:34.068882 134083346794304 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10. I0418 22:27:34.637984 134083346794304 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0418 22:27:34.638089 134083346794304 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 I0418 22:27:34.690617 134083346794304 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.055132s I0418 22:27:34.690785 134083346794304 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 22.134 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06969857215881348 s) (per-host) I0418 22:27:34.690834 134083346794304 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.069758s (batch_requests_ready=0.005221s, total_serialization_initiated=0.064470s, others=0.000068s) I0418 22:27:34.690950 134083346794304 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.074235s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.074219) I0418 22:27:34.691679 134083346794304 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10. I0418 22:27:34.691966 133957102315264 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 22:47:34.691944 I0418 22:27:35.129730 134083346794304 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0418 22:27:35.130143 133953914660608 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0418 22:27:35.130305 134083346794304 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776551250.0940945, 'wait_for_prev_duration_secs': 3.971163272857666, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776551254.066846, 'checkpointer_blocking_duration_secs': 0.6252150535583496, 'get_old_steps_start_time': 1776551254.6920843, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776551250.0922213, 'checkpoint_manager_blocking_duration_secs': 5.038040399551392} I0418 22:27:35.130522 134083346794304 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0418 22:27:35.131226 134083346794304 metric_logger.py:196] completed step: 9, seconds: 0.157, TFLOP/s/device: 86.665, Tokens/s/device: 13063.057, total_weights: 65536, loss: 10.656, lm_loss: 10.656, perplexity: 42467.320 I0418 22:27:35.142580 134083346794304 metric_logger.py:196] completed step: 10, seconds: 0.157, TFLOP/s/device: 86.525, Tokens/s/device: 13041.928, total_weights: 65536, loss: 10.636, lm_loss: 10.636, perplexity: 41601.941 I0418 22:27:35.295506 134083346794304 metric_logger.py:196] completed step: 11, seconds: 5.354, TFLOP/s/device: 2.538, Tokens/s/device: 382.494, total_weights: 65536, loss: 10.618, lm_loss: 10.618, perplexity: 40849.570 I0418 22:27:35.452214 134083346794304 metric_logger.py:196] completed step: 12, seconds: 0.010, TFLOP/s/device: 1382.914, Tokens/s/device: 208447.837, total_weights: 65536, loss: 10.602, lm_loss: 10.602, perplexity: 40203.926 I0418 22:27:35.609091 134083346794304 metric_logger.py:196] completed step: 13, seconds: 0.154, TFLOP/s/device: 88.050, Tokens/s/device: 13271.813, total_weights: 65536, loss: 10.588, lm_loss: 10.588, perplexity: 39652.145 I0418 22:27:35.765988 134083346794304 metric_logger.py:196] completed step: 14, seconds: 0.157, TFLOP/s/device: 86.689, Tokens/s/device: 13066.725, total_weights: 65536, loss: 10.577, lm_loss: 10.577, perplexity: 39212.879 I0418 22:27:36.340371 134083346794304 metric_logger.py:196] completed step: 15, seconds: 0.157, TFLOP/s/device: 86.684, Tokens/s/device: 13065.891, total_weights: 65536, loss: 10.568, lm_loss: 10.568, perplexity: 38879.223 I0418 22:27:36.497265 134083346794304 metric_logger.py:196] completed step: 16, seconds: 0.568, TFLOP/s/device: 23.905, Tokens/s/device: 3603.242, total_weights: 65536, loss: 10.562, lm_loss: 10.562, perplexity: 38619.105 I0418 22:27:36.653983 134083346794304 metric_logger.py:196] completed step: 17, seconds: 0.163, TFLOP/s/device: 83.392, Tokens/s/device: 12569.738, total_weights: 65536, loss: 10.556, lm_loss: 10.556, perplexity: 38395.816 I0418 22:27:36.810895 134083346794304 metric_logger.py:196] completed step: 18, seconds: 0.157, TFLOP/s/device: 86.602, Tokens/s/device: 13053.649, total_weights: 65536, loss: 10.552, lm_loss: 10.552, perplexity: 38236.652 I0418 22:27:36.969659 134083346794304 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0418 22:27:36.970897 134083346794304 checkpointing.py:798] Waited 0.0012493133544921875 seconds for step 19 to finish before starting checkpointing. I0418 22:27:36.973015 134083346794304 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 22:27:41.951358 133954417960704 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_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10/items/array_metadatas/process_6 I0418 22:27:57.982889 133957102315264 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 67.622 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 23.36176371574402 s) (per-host) I0418 22:27:57.983024 133957102315264 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 23.290984s. I0418 22:28:06.455292 133957102315264 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 31.763237s. I0418 22:28:06.455680 133953914660608 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 22:28:06.455796 133953914660608 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0418 22:28:06.455845 133953914660608 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0418 22:28:06.457378 133953914660608 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0418 22:28:06.457568 134083346794304 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. W0418 22:28:06.457697 134083346794304 checkpoint_manager.py:1452] Waiting for previous save to complete took 29.484685 seconds. If this number is high, consider checkpointing less frequently. I0418 22:28:06.459339 134083346794304 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0418 22:28:06.461395 134083346794304 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19. I0418 22:28:07.614104 134083346794304 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0418 22:28:07.614206 134083346794304 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 I0418 22:28:07.665153 134083346794304 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.053505s I0418 22:28:07.665335 134083346794304 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 22.843 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0675356388092041 s) (per-host) I0418 22:28:07.665387 134083346794304 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.067597s (batch_requests_ready=0.005176s, total_serialization_initiated=0.062355s, others=0.000066s) I0418 22:28:07.665508 134083346794304 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.071940s (all_items=0.000012s, per_item={'items': '0.00001192'}, temp_paths=0.071928) I0418 22:28:07.666189 134083346794304 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.21 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19. I0418 22:28:07.666738 133953914660608 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 22:48:07.666702 I0418 22:28:07.668731 134083346794304 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0418 22:28:07.669011 133957132056320 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0418 22:28:07.669169 134083346794304 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776551256.9729857, 'wait_for_prev_duration_secs': 29.48468518257141, 'time_between_consecutive_saves_sec': 2.9079782962799072, 'checkpointer_blocking_start_time': 1776551286.4593792, 'checkpointer_blocking_duration_secs': 1.207503318786621, 'get_old_steps_start_time': 1776551287.6669014, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776551256.9712014, 'checkpoint_manager_blocking_duration_secs': 30.69793677330017} I0418 22:28:07.669309 134083346794304 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0418 22:28:07.669356 134083346794304 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 22:28:14.742507 133962462631680 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_linen_converter_and_sharding_tools_20260418_204152/nnx_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19/items/array_metadatas/process_6 I0418 22:28:29.933605 133953914660608 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 70.728 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 22.33574342727661 s) (per-host) I0418 22:28:29.933743 133953914660608 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 22.266891s. I0418 22:28:38.154952 133953914660608 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 30.488083s. I0418 22:28:38.155262 133957132056320 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 22:28:38.155384 133957132056320 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0418 22:28:38.155425 133957132056320 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0418 22:28:38.157630 133957132056320 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0418 22:28:38.157801 134083346794304 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. I0418 22:28:38.157982 134083346794304 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 22:28:38.158906 134083346794304 metric_logger.py:196] completed step: 19, seconds: 0.157, TFLOP/s/device: 86.679, Tokens/s/device: 13065.141, 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: Sat Apr 18 22:28:48 UTC 2026 EXIT_CODE=0