[2026-01-26 21:45:41,287] [WARNING] [axolotl.utils.trainer.prepare_optim_env:658] [PID:58141] P2P support not detected, setting `NCCL_P2P_DISABLE=1` [2026-01-26 21:45:41,445] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:58141] baseline 0.000GB () [2026-01-26 21:45:41,445] [INFO] [axolotl.cli.config.load_cfg:259] [PID:58141] config: { "accelerator_config": { "dispatch_batches": false, "split_batches": false }, "activation_offloading": false, "adapter": "lora", "axolotl_config_path": "./qlora-32b-part2.yaml", "base_model": "Guilherme34/secretmodel-indevelopment-full-part1", "base_model_config": "Guilherme34/secretmodel-indevelopment-full-part1", "batch_size": 2, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_89", "fp8": false, "n_gpu": 2, "n_node": 1 }, "context_parallel_size": 1, "dataloader_num_workers": 2, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_num_proc": 54, "dataset_prepared_path": "last_run_prepared", "ddp": true, "device": "cuda:0", "device_map": { "": 0 }, "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "env_capabilities": { "torch_version": "2.9.1" }, "eval_batch_size": 1, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": true, "eval_table_size": 0, "evals_per_epoch": 1, "experimental_skip_move_to_device": true, "flash_attention": true, "fp16": false, "fsdp": [ "full_shard", "auto_wrap" ], "fsdp_config": { "auto_wrap_policy": "TRANSFORMER_BASED_WRAP", "cpu_ram_efficient_loading": true, "offload_params": true, "state_dict_type": "FULL_STATE_DICT", "sync_module_states": true, "transformer_layer_cls_to_wrap": "Glm4MoeLiteDecoderLayer", "use_orig_params": false }, "gradient_accumulation_steps": 1, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": true }, "include_tkps": true, "learning_rate": 0.0002, "lisa_layers_attribute": "model.layers", "load_best_model_at_end": false, "load_in_4bit": false, "load_in_8bit": false, "local_rank": 0, "logging_steps": 1, "lora_alpha": 48, "lora_dropout": 0.05, "lora_r": 24, "lora_target_modules": [ "gate_proj", "down_proj", "up_proj", "q_proj", "v_proj", "k_proj", "o_proj" ], "loraplus_lr_embedding": 1e-06, "loss_watchdog_patience": 3, "loss_watchdog_threshold": 5.0, "lr_scheduler": "cosine", "max_steps": 5680, "mean_resizing_embeddings": false, "micro_batch_size": 1, "model_config_type": "glm4_moe_lite", "num_epochs": 1.0, "optimizer": "adamw_torch_fused", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "./outputs/qlora-out", "pad_to_sequence_len": true, "pretrain_multipack_attn": true, "pretraining_dataset": [ { "path": "Guilherme34/best-dataset-glm47flash", "split": "train", "text_column": "text", "trust_remote_code": false, "type": "pretrain" } ], "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "ray_num_workers": 1, "resources_per_worker": { "GPU": 1 }, "sample_packing": true, "sample_packing_bin_size": 200, "sample_packing_group_size": 100000, "save_only_model": false, "save_safetensors": true, "save_steps": 1000, "sequence_len": 4096, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": false, "special_tokens": { "pad_token": "<|endoftext|>" }, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tf32": false, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "Guilherme34/secretmodel-indevelopment-full-part1", "tokenizer_save_jinja_files": true, "torch_dtype": "torch.bfloat16", "train_on_inputs": false, "trl": { "log_completions": false, "mask_truncated_completions": false, "ref_model_mixup_alpha": 0.9, "ref_model_sync_steps": 64, "scale_rewards": true, "sync_ref_model": false, "use_vllm": false, "vllm_server_host": "0.0.0.0", "vllm_server_port": 8000 }, "use_otel_metrics": false, "use_ray": false, "val_set_size": 0.0, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "warmup_ratio": 0.1, "weight_decay": 0.0, "world_size": 2 } [2026-01-26 21:45:41,445] [WARNING] [axolotl.cli.checks.check_accelerate_default_config:19] [PID:58141] accelerate config file found at /root/.cache/huggingface/accelerate/default_config.yaml. This can lead to unexpected errors [2026-01-26 21:45:43,488] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:58141] EOS: 154820 / <|endoftext|> [2026-01-26 21:45:43,488] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:58141] BOS: None / None [2026-01-26 21:45:43,488] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:58141] PAD: 154820 / <|endoftext|> [2026-01-26 21:45:43,489] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:58141] UNK: None / None [2026-01-26 21:45:45,394] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:58141] loading tokenizer... Guilherme34/secretmodel-indevelopment-full-part1 [2026-01-26 21:45:47,156] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:58141] EOS: 154820 / <|endoftext|> [2026-01-26 21:45:47,156] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:58141] BOS: None / None [2026-01-26 21:45:47,156] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:58141] PAD: 154820 / <|endoftext|> [2026-01-26 21:45:47,156] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:58141] UNK: None / None [2026-01-26 21:45:47,156] [DEBUG] [axolotl.train.setup_model_and_tokenizer:82] [PID:58141] Loading model [2026-01-26 21:45:47,280] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:58141] Patched Trainer.evaluation_loop with nanmean loss calculation [2026-01-26 21:45:47,281] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:58141] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2026-01-26 21:45:47,281] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:345] [PID:58141] Applying multipack dataloader patch for sample packing... Loading weights: 0%| | 0/751 [00:00:39] [PID:58141] Skipping import of cpp extensions due to incompatible torch version 2.9.1+cu130 for torchao version 0.13.0 [2026-01-26 21:48:43,934] [WARNING] [torchao.:39] [PID:58142] Skipping import of cpp extensions due to incompatible torch version 2.9.1+cu130 for torchao version 0.13.0 trainable params: 20,840,448 || all params: 29,964,231,424 || trainable%: 0.0696 [2026-01-26 21:48:44,155] [DEBUG] [axolotl.loaders.model.log_gpu_memory_usage:127] [PID:58141] after adapters 0.000GB () [2026-01-26 21:48:48,350] [WARNING] [accelerate.utils.dataclasses.__post_init__:1962] [PID:58142] sharding_strategy is deprecated in favor of reshard_after_forward. This will be removed in a future version of Accelerate. [2026-01-26 21:48:48,350] [WARNING] [accelerate.utils.dataclasses.__post_init__:1962] [PID:58141] sharding_strategy is deprecated in favor of reshard_after_forward. This will be removed in a future version of Accelerate. [2026-01-26 21:48:48,893] [INFO] [axolotl.train.save_initial_configs:413] [PID:58141] Pre-saving adapter config to ./outputs/qlora-out... [2026-01-26 21:48:48,894] [INFO] [axolotl.train.save_initial_configs:417] [PID:58141] Pre-saving tokenizer to ./outputs/qlora-out... [2026-01-26 21:48:49,037] [INFO] [axolotl.train.save_initial_configs:422] [PID:58141] Pre-saving model config to ./outputs/qlora-out... [2026-01-26 21:48:49,048] [INFO] [axolotl.train.execute_training:212] [PID:58141] Starting trainer... 0%| | 0/5680 [00:00