XPK Start: Thu Apr 23 12:54:37 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 12:55:17.061731: 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 12:55:17.271543 136540902844224 max_utils.py:273] Attempting to initialize the jax distributed system... I0423 12:55:26.313263 136540902844224 distributed.py:149] Starting JAX distributed service on [::]:8482 I0423 12:55:26.315655 136540902844224 distributed.py:172] Connecting to JAX distributed service on mt-04-int8-u17qx-slice-job-0-0.mt-04-int8-u17qx:8482 I0423 12:56:51.289558 136540902844224 max_utils.py:284] Jax distributed system initialized! I0423 12:56:57.363053 136540902844224 max_utils.py:800] System Information: Jax Version: 0.9.2 I0423 12:56:57.363158 136540902844224 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0423 12:56:57.363199 136540902844224 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 12:56:57.363234 136540902844224 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 12:56:58.051234 136540902844224 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 12:56:58.051836 136540902844224 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 12:56:58.052133 136540902844224 checkpointing.py:688] Setting up checkpoint logger... I0423 12:56:58.052188 136540902844224 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 12:56:58.052229 136540902844224 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0423 12:56:58.052558 136540902844224 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 0x7c2e305481a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 12:57:00.961781 136540902844224 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 12:57:00.962012 136540902844224 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c1a1c71e360>}, handler_registry=None I0423 12:57:00.962261 136540902844224 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c1a1c71e360>` 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 12:57:00.962311 136540902844224 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c1a1c721e50>` 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 12:57:00.962347 136540902844224 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 0x7c1a1c71e360>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c1a1c71e360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c1a1c721e50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c1a1c721e50>}). I0423 12:57:00.962696 136540902844224 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0423 12:57:00.962766 136540902844224 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c1a044d9d00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0423 12:57:01.918324 136540902844224 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints I0423 12:57:02.373566 136540902844224 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_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/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c1a1c71c680> I0423 12:57:02.373769 136540902844224 checkpointing.py:302] Checkpoint manager created! I0423 12:57:03.503513 136540902844224 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 12:57:03.503635 136540902844224 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 12:57:03.885247 136540902844224 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 12:57:03.885338 136540902844224 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 12:57:03.902776 136540902844224 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 12:57:03.902837 136540902844224 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 12:57:03.970505 136540902844224 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 12:57:03.970608 136540902844224 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 12:57:03.988056 136540902844224 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 12:57:03.988124 136540902844224 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 12:57:04.005390 136540902844224 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 12:57:04.005450 136540902844224 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 12:57:04.031102 136540902844224 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 12:57:04.031172 136540902844224 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 12:57:04.100137 136540902844224 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 12:57:04.100224 136540902844224 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 12:57:04.626944 136540902844224 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 12:57:04.627056 136540902844224 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0423 12:57:07.116345 136540902844224 maxtext_utils.py:1880] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0423 12:57:07.116472 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 12:57:07.116525 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 12:57:07.116593 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0423 12:57:07.116652 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0423 12:57:07.116692 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0423 12:57:07.116742 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 12:57:07.116793 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0423 12:57:07.116832 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0423 12:57:07.116867 136540902844224 maxtext_utils.py:1880] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 12:57:07.116915 136540902844224 maxtext_utils.py:1880] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 12:57:07.116963 136540902844224 maxtext_utils.py:1880] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 12:57:08.539877 136540902844224 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0423 12:57:08.539973 136540902844224 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0423 12:57:08.557307 136540902844224 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0423 12:57:08.557377 136540902844224 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0423 12:57:21.940917 136540902844224 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. I0423 12:57:21.948770 136540902844224 metric_logger.py:301] number parameters: 1.104 billion I0423 12:57:37.596954 136540902844224 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0423 12:57:37.837825 136540902844224 checkpointing.py:798] Waited 0.2408466339111328 seconds for step 0 to finish before starting checkpointing. I0423 12:57:37.840336 136540902844224 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 12:57:37.842259 136540902844224 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0423 12:57:37.843723 136540902844224 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0. I0423 12:57:38.569670 136540902844224 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 12:57:38.584203 136540902844224 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0423 12:57:38.584271 136540902844224 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 12:57:38.859876 136540902844224 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.276590s I0423 12:57:38.860693 136540902844224 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.453 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28286075592041016 s) (per-host) I0423 12:57:38.860764 136540902844224 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.282949s (batch_requests_ready=0.002081s, total_serialization_initiated=0.280135s, others=0.000733s) I0423 12:57:38.860841 136540902844224 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287109s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.287092) I0423 12:57:38.861680 136540902844224 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0. I0423 12:57:38.862014 136409586583296 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 13:17:38.861976 I0423 12:57:38.865974 136409051576064 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0 I0423 12:57:38.872647 136540902844224 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 12:57:38.872932 136406100924160 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 12:57:38.873114 136540902844224 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776949057.8403168, 'wait_for_prev_duration_secs': 6.079673767089844e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776949057.8422976, 'checkpointer_blocking_duration_secs': 1.0198726654052734, 'get_old_steps_start_time': 1776949058.8621955, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776949057.8383317, 'checkpoint_manager_blocking_duration_secs': 1.0347309112548828} I0423 12:57:38.873218 136540902844224 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0423 12:57:38.873285 136540902844224 max_utils.py:750] Memstats: After params initialized: I0423 12:57:38.873343 136540902844224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0423 12:57:38.873378 136540902844224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0423 12:57:38.873405 136540902844224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0423 12:57:38.873429 136540902844224 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0423 12:57:39.229451 136540902844224 metric_logger.py:196] completed step: 0, seconds: 15.339, TFLOP/s/device: 0.886, Tokens/s/device: 133.512, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277 I0423 12:57:39.232596 136540902844224 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/tensorboard/' I0423 12:57:39.639229 136540902844224 metric_logger.py:196] completed step: 1, seconds: 1.630, TFLOP/s/device: 8.333, Tokens/s/device: 1256.083, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277 I0423 12:57:39.866645 136540902844224 metric_logger.py:196] completed step: 2, seconds: 0.411, TFLOP/s/device: 33.092, Tokens/s/device: 4988.054, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28634.820 I0423 12:57:40.096156 136540902844224 metric_logger.py:196] completed step: 3, seconds: 0.010, TFLOP/s/device: 1402.180, Tokens/s/device: 211351.909, total_weights: 65536, loss: 9.731, lm_loss: 9.731, perplexity: 16827.199 I0423 12:57:40.662059 136408536471296 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/items I0423 12:57:41.035623 136406092531456 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776949060412193153, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776949060412193153, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0423 12:57:41.825618 2961 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 12:57:43.958224 136408508462848 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/items/array_metadatas/process_0 I0423 12:58:10.027926 136540902844224 metric_logger.py:196] completed step: 4, seconds: 0.228, TFLOP/s/device: 59.567, Tokens/s/device: 8978.636, total_weights: 65536, loss: 9.272, lm_loss: 9.272, perplexity: 10638.531 I0423 12:58:10.036983 136540902844224 metric_logger.py:196] completed step: 5, seconds: 0.230, TFLOP/s/device: 59.171, Tokens/s/device: 8918.928, total_weights: 65536, loss: 8.887, lm_loss: 8.887, perplexity: 7233.941 I0423 12:58:10.264823 136540902844224 metric_logger.py:196] completed step: 6, seconds: 29.930, TFLOP/s/device: 0.454, Tokens/s/device: 68.425, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5365.468 I0423 12:58:10.494349 136540902844224 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2054.609, Tokens/s/device: 309693.029, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4358.779 I0423 12:58:10.723752 136540902844224 metric_logger.py:196] completed step: 8, seconds: 0.229, TFLOP/s/device: 59.371, Tokens/s/device: 8949.093, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.453 I0423 12:58:10.952481 136540902844224 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0423 12:58:10.953175 136540902844224 checkpointing.py:798] Waited 0.0007147789001464844 seconds for step 9 to finish before starting checkpointing. I0423 12:58:10.955184 136540902844224 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 12:58:16.396847 136406109316864 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.579641s (commit=33.489524s, array_metadata_write=1.090118s) I0423 12:58:16.398036 136409586583296 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 41.766 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.82017374038696 s) (per-host) I0423 12:58:16.398178 136409586583296 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.536052s. I0423 12:58:17.073736 136409586583296 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776949060412193153, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0423 12:58:17.791975 136409586583296 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000269s. I0423 12:58:17.833684 136406092531456 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776949060412193153, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0423 12:58:18.284112 136409586583296 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0423 12:58:19.242356 136409586583296 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.040234s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/items I0423 12:58:19.244231 136409586583296 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0/items I0423 12:58:19.774308 136409586583296 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0 I0423 12:58:21.191758 136409586583296 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0`. I0423 12:58:21.192614 136409586583296 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 43.35 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/0 I0423 12:58:21.194213 136409586583296 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 42.332084s. I0423 12:58:21.194421 136406100924160 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 12:58:21.194546 136406100924160 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 12:58:21.194628 136406100924160 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 12:58:21.197252 136406100924160 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 12:58:21.197439 136540902844224 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0423 12:58:21.197573 136540902844224 checkpoint_manager.py:1452] Waiting for previous save to complete took 10.242389 seconds. If this number is high, consider checkpointing less frequently. I0423 12:58:21.199478 136540902844224 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0423 12:58:21.201542 136540902844224 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9. I0423 12:58:21.924302 136540902844224 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0423 12:58:21.924393 136540902844224 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 12:58:21.949836 136540902844224 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026186s I0423 12:58:21.950636 136540902844224 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 50.282 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030678749084472656 s) (per-host) I0423 12:58:21.950714 136540902844224 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030775s (batch_requests_ready=0.001845s, total_serialization_initiated=0.028200s, others=0.000729s) I0423 12:58:21.950799 136540902844224 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034617s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.034600) I0423 12:58:21.951561 136540902844224 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9. I0423 12:58:21.951873 136406100924160 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 13:18:21.951848 I0423 12:58:21.961725 136540902844224 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 12:58:21.961987 136408010315520 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 12:58:21.962153 136540902844224 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776949090.9551513, 'wait_for_prev_duration_secs': 10.24238896369934, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776949101.1995184, 'checkpointer_blocking_duration_secs': 0.7524504661560059, 'get_old_steps_start_time': 1776949101.9519928, 'get_old_steps_duration_secs': 3.0994415283203125e-05, 'checkpoint_manager_blocking_start_time': 1776949090.9534464, 'checkpoint_manager_blocking_duration_secs': 11.008668184280396} I0423 12:58:21.962262 136540902844224 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0423 12:58:21.962308 136540902844224 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 12:58:22.148499 136409586583296 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9 I0423 12:58:24.361142 136409051576064 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9/items I0423 12:58:27.261982 136408027100928 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9/items/array_metadatas/process_0 I0423 12:59:07.335441 136408018708224 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.839662s (commit=40.576030s, array_metadata_write=1.263632s) I0423 12:59:07.337463 136406100924160 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.779 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.41750907897949 s) (per-host) I0423 12:59:07.337528 136406100924160 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.385581s. I0423 12:59:08.720287 136406100924160 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000271s. I0423 12:59:09.383520 136406100924160 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0423 12:59:10.317924 136406100924160 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.171509s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9/items I0423 12:59:10.319555 136406100924160 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9/items I0423 12:59:10.850660 136406100924160 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9 I0423 12:59:12.696210 136406100924160 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9`. I0423 12:59:12.697055 136406100924160 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.50 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/linen_xpk_feat_nnx_post_train_fixes_20260423_124541_04_int8/checkpoints/9 I0423 12:59:12.698177 136406100924160 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.746227s. I0423 12:59:12.698375 136408010315520 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 12:59:12.698496 136408010315520 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 12:59:12.698559 136408010315520 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 12:59:12.700454 136408010315520 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 12:59:12.700600 136540902844224 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0423 12:59:12.700699 136540902844224 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 12:59:12.701705 136540902844224 metric_logger.py:196] completed step: 9, seconds: 0.229, TFLOP/s/device: 59.228, Tokens/s/device: 8927.443, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3552.333 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 23 12:59:23 UTC 2026 EXIT_CODE=0