[2025-11-13 03:45:45,108] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:1248082] bf16 support detected, enabling for this configuration. [2025-11-13 03:45:45,108] [WARNING] [py.warnings._showwarnmsg:112] [PID:1248082] /root/miniforge3/envs/axolotl/lib/python3.12/site-packages/torch/backends/__init__.py:46: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.) self.setter(val) [2025-11-13 03:45:45,221] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:1248082] baseline 0.000GB () [2025-11-13 03:45:45,221] [INFO] [axolotl.cli.config.load_cfg:248] [PID:1248082] config: { "activation_offloading": false, "axolotl_config_path": "171-llama3.2-3b-v2.1-153sft-dpo-2e7.axolotl.1xMI300.yaml", "base_model": "shisa-ai/153-llama3.2-3b-v2.1-sft", "base_model_config": "shisa-ai/153-llama3.2-3b-v2.1-sft", "batch_size": 64, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_94", "fp8": false, "n_gpu": 1, "n_node": 1 }, "chat_template": "tokenizer_default", "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_num_proc": 20, "dataset_prepared_path": "cached_formatted_dataset/171", "datasets": [ { "path": "dpo.shisa-v2.1.jsonl", "split": "train", "type": "chat_template.argilla_chat" } ], "ddp": false, "deepspeed": { "bf16": { "enabled": true }, "gradient_accumulation_steps": "auto", "gradient_clipping": "auto", "train_batch_size": "auto", "train_micro_batch_size_per_gpu": "auto", "wall_clock_breakdown": false, "zero_optimization": { "contiguous_gradients": true, "overlap_comm": true, "reduce_bucket_size": "auto", "stage": 3, "stage3_gather_16bit_weights_on_model_save": true, "stage3_max_live_parameters": 0, "stage3_max_reuse_distance": 0, "stage3_param_persistence_threshold": "auto", "stage3_prefetch_bucket_size": "auto", "sub_group_size": 0 } }, "device": "cuda:0", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "env_capabilities": { "torch_version": "2.9.0" }, "eval_batch_size": 8, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_table_size": 0, "evals_per_epoch": 0, "experimental_skip_move_to_device": true, "flash_attention": true, "fp16": false, "gradient_accumulation_steps": 8, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": false }, "group_by_length": false, "include_tkps": true, "is_llama_derived_model": true, "learning_rate": 2e-07, "liger_fused_linear_cross_entropy": true, "liger_glu_activation": true, "liger_rms_norm": true, "liger_rope": true, "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_dropout": 0.0, "loraplus_lr_embedding": 1e-06, "lr_scheduler": "constant_with_warmup", "mean_resizing_embeddings": false, "micro_batch_size": 8, "model_config_type": "llama", "neftune_noise_alpha": 5.0, "num_epochs": 1.0, "optimizer": "adamw_torch_8bit", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "/data/outputs/171-llama3.2-3b-v2.1-153sft-dpo-2e7", "pad_to_sequence_len": false, "plugins": [ "axolotl.integrations.liger.LigerPlugin" ], "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "ray_num_workers": 1, "remove_unused_columns": false, "resources_per_worker": { "GPU": 1 }, "rl": "dpo", "rl_beta": 0.1, "sample_packing": false, "sample_packing_bin_size": 200, "sample_packing_group_size": 100000, "save_only_model": false, "save_safetensors": true, "save_total_limit": 1, "saves_per_epoch": 0, "sequence_len": 2048, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": true, "special_tokens": { "pad_token": "<|end_of_text|>" }, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tf32": false, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "shisa-ai/153-llama3.2-3b-v2.1-sft", "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, "use_wandb": true, "val_set_size": 0.0, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "wandb_entity": "augmxnt", "wandb_name": "171-llama3.2-3b-v2.1-153sft-dpo-2e7", "wandb_project": "shisa-v2.1", "warmup_ratio": 0.03, "weight_decay": 0.0001, "world_size": 1 } [2025-11-13 03:45:45,895] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:1248082] EOS: 128009 / <|eot_id|> [2025-11-13 03:45:45,895] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:1248082] BOS: 128000 / <|begin_of_text|> [2025-11-13 03:45:45,895] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1248082] PAD: 128001 / <|end_of_text|> [2025-11-13 03:45:45,895] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1248082] UNK: None / None [2025-11-13 03:45:45,896] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:1248082] Unable to find prepared dataset in cached_formatted_dataset/171/180e7b2d86ffa9c542c8863eed490da8 [2025-11-13 03:45:46,750] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:1248082] EOS: 128009 / <|eot_id|> [2025-11-13 03:45:46,750] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:1248082] BOS: 128000 / <|begin_of_text|> [2025-11-13 03:45:46,750] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1248082] PAD: 128001 / <|end_of_text|> [2025-11-13 03:45:46,750] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1248082] UNK: None / None [2025-11-13 03:45:46,919] [DEBUG] [axolotl.train.setup_model_and_tokenizer:65] [PID:1248082] Loading tokenizer... shisa-ai/153-llama3.2-3b-v2.1-sft [2025-11-13 03:45:47,615] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:1248082] EOS: 128009 / <|eot_id|> [2025-11-13 03:45:47,615] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:1248082] BOS: 128000 / <|begin_of_text|> [2025-11-13 03:45:47,615] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1248082] PAD: 128001 / <|end_of_text|> [2025-11-13 03:45:47,615] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1248082] UNK: None / None [2025-11-13 03:45:47,615] [DEBUG] [axolotl.train.setup_model_and_tokenizer:74] [PID:1248082] Loading model [2025-11-13 03:45:47,697] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:1248082] Patched Trainer.evaluation_loop with nanmean loss calculation [2025-11-13 03:45:47,698] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:1248082] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2025-11-13 03:45:47,712] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:71] [PID:1248082] Applying LIGER to llama with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': True, 'rms_norm': True, 'swiglu': True} [2025-11-13 03:45:47,724] [WARNING] [py.warnings._showwarnmsg:112] [PID:1248082] /root/miniforge3/envs/axolotl/lib/python3.12/site-packages/deepspeed/runtime/zero/partition_parameters.py:255: UserWarning: expandable_segments not supported on this platform (Triggered internally at /pytorch/c10/hip/HIPAllocatorConfig.h:36.) tensor: Tensor = fn(*args, **kwargs) Loading checkpoint shards: 0%| | 0/2 [00:00