[2026-03-23 11:12:13,566] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:14407] baseline 0.000GB () [2026-03-23 11:12:13,568] [INFO] [axolotl.cli.config.load_cfg:341] [PID:14407] config: { "activation_offloading": false, "adapter": "lora", "auto_resume_from_checkpoints": true, "axolotl_config_path": "llama_train.yml", "base_model": "meta-llama/Llama-3.2-3B-Instruct", "base_model_config": "meta-llama/Llama-3.2-3B-Instruct", "batch_size": 4, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_89", "fp8": false, "n_gpu": 1, "n_node": 1, "tf32": true }, "chat_template": "tokenizer_default", "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_num_proc": 120, "datasets": [ { "message_property_mappings": { "content": "content", "role": "role" }, "path": "psychopenguin/indian_legal_dataset_qna", "split": "train", "trust_remote_code": false, "type": "alpaca" } ], "ddp": false, "device": "cuda:0", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "eaft_alpha": 1.0, "eaft_k": 20, "early_stopping_patience": 6, "env_capabilities": { "torch_version": "2.9.1" }, "eval_batch_size": 4, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": true, "eval_steps": 100, "eval_strategy": "steps", "eval_table_size": 0, "experimental_skip_move_to_device": true, "flash_attention": true, "flash_attn_cross_entropy": true, "fp16": false, "generate_samples": false, "generation_do_sample": true, "generation_max_new_tokens": 50, "generation_prompt_ratio": 0.5, "generation_temperature": 0.7, "gradient_accumulation_steps": 1, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": true }, "hf_use_auth_token": true, "hub_model_id": "psychopenguin/legal_llama3.2-3b-instruct", "include_tkps": true, "is_falcon_derived_model": false, "is_llama_derived_model": true, "is_mistral_derived_model": false, "layer_offloading": false, "learning_rate": 0.0002, "lisa_layers_attribute": "model.layers", "load_best_model_at_end": true, "load_in_4bit": false, "load_in_8bit": false, "local_rank": 0, "lora_alpha": 32, "lora_dropout": 0.05, "lora_r": 16, "lora_target_linear": true, "loraplus_lr_embedding": 1e-06, "lr_scheduler": "cosine", "mean_resizing_embeddings": false, "merge_lora": true, "micro_batch_size": 4, "model_config_type": "llama", "neftune_noise_alpha": 5.0, "num_epochs": 5.0, "num_generation_samples": 3, "optimizer": "paged_adamw_8bit", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "./final_model", "pad_to_sequence_len": true, "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "quantize_moe_experts": 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": 100, "save_total_limit": 3, "seed": 9, "sequence_len": 2048, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": false, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tf32": true, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "meta-llama/Llama-3.2-3B-Instruct", "tokenizer_save_jinja_files": true, "tokenizer_type": "AutoTokenizer", "tokenizer_use_fast": true, "torch_dtype": "torch.bfloat16", "train_on_inputs": false, "trl": { "async_prefetch": false, "log_completions": false, "mask_truncated_completions": false, "ref_model_mixup_alpha": 0.9, "ref_model_sync_steps": 64, "replay_buffer_size": 0, "replay_recompute_logps": true, "reroll_max_groups": 1, "reroll_start_fraction": 1.0, "reward_num_workers": 1, "scale_rewards": true, "skip_zero_advantage_batches": true, "sync_ref_model": false, "use_data_producer": false, "use_vllm": false, "vllm_lora_sync": false, "vllm_server_host": "0.0.0.0", "vllm_server_port": 8000 }, "trust_remote_code": true, "type_of_model": "AutoModelForCausalLM", "use_otel_metrics": false, "use_ray": false, "use_wandb": true, "val_set_size": 0.05, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "wandb_name": "legal-lm", "wandb_project": "major_project", "weight_decay": 0.0, "world_size": 1 } [2026-03-23 11:12:17,084] [WARNING] [axolotl.loaders.tokenizer.load_tokenizer:227] [PID:14407] Tokenizer does not have a pad_token, falling back to eos_token: <|eot_id|> [2026-03-23 11:12:17,085] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:307] [PID:14407] EOS: 128009 / <|eot_id|> [2026-03-23 11:12:17,086] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:308] [PID:14407] BOS: 128000 / <|begin_of_text|> [2026-03-23 11:12:17,086] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:309] [PID:14407] PAD: 128009 / <|eot_id|> [2026-03-23 11:12:17,086] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:310] [PID:14407] UNK: None / None [2026-03-23 11:12:17,087] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:480] [PID:14407] Unable to find prepared dataset in last_run_prepared/bbd35feda38ef811b8a988570a6ea262 [2026-03-23 11:12:17,087] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:14407] Loading raw datasets... [2026-03-23 11:12:17,088] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:14407] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`. Downloading (incomplete total...): 0.00B [00:00, ?B/s] Fetching 0 files: 0it [00:00, ?it/s] Fetching 0 files: 0it [00:00, ?it/s] Download complete: : 0.00B [00:00, ?B/s] Download complete: : 0.00B [00:02, ?B/s] [2026-03-23 11:12:21,343] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:14407] Loading dataset: psychopenguin/indian_legal_dataset_qna with base_type: alpaca and prompt_style: None [2026-03-23 11:12:21,962] [INFO] [axolotl.utils.data.utils._log_dataset_stats:212] [PID:14407] min_input_len: 25 [2026-03-23 11:12:21,963] [INFO] [axolotl.utils.data.utils._log_dataset_stats:213] [PID:14407] max_input_len: 2092 [2026-03-23 11:12:23,140] [INFO] [axolotl.utils.data.utils._drop_outside_range:306] [PID:14407] Dropped 3 sequences outside valid range ([None, 2048]) Saving the dataset (0/120 shards): 0%| | 0/37045 [00:00 [2026-03-23 11:12:49,878] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:307] [PID:14407] EOS: 128009 / <|eot_id|> [2026-03-23 11:12:49,878] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:308] [PID:14407] BOS: 128000 / <|begin_of_text|> [2026-03-23 11:12:49,878] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:309] [PID:14407] PAD: 128009 / <|eot_id|> [2026-03-23 11:12:49,879] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:310] [PID:14407] UNK: None / None [2026-03-23 11:12:49,879] [DEBUG] [axolotl.train.setup_model_and_tokenizer:81] [PID:14407] Loading model [2026-03-23 11:12:50,154] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:91] [PID:14407] Patched Trainer.evaluation_loop with nanmean loss calculation [2026-03-23 11:12:50,157] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:142] [PID:14407] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2026-03-23 11:12:50,159] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:402] [PID:14407] Applying multipack dataloader patch for sample packing... [2026-03-23 11:12:50,293] [INFO] [axolotl.monkeypatch.llama_attn_hijack_flash.patch_fa_llama_cross_entropy:69] [PID:14407] patching transformers.loss.loss_utils.fixed_cross_entropy with flash_attn.ops.triton.cross_entropy Loading weights: 0%| | 0/254 [00:00