| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file tokenizer.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/tokenizer.json | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file tokenizer.model from cache at None | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file added_tokens.json from cache at None | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file special_tokens_map.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/special_tokens_map.json | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file tokenizer_config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/tokenizer_config.json | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2060 >> loading file chat_template.jinja from cache at None | |
| [INFO|2025-04-17 02:09:41] tokenization_utils_base.py:2323 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. | |
| [INFO|2025-04-17 02:09:45] configuration_utils.py:693 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/config.json | |
| [INFO|2025-04-17 02:09:45] configuration_utils.py:765 >> Model config LlamaConfig { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 8, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "factor": 32.0, | |
| "high_freq_factor": 4.0, | |
| "low_freq_factor": 1.0, | |
| "original_max_position_embeddings": 8192, | |
| "rope_type": "llama3" | |
| }, | |
| "rope_theta": 500000.0, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.51.1", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file tokenizer.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/tokenizer.json | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file tokenizer.model from cache at None | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file added_tokens.json from cache at None | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file special_tokens_map.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/special_tokens_map.json | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file tokenizer_config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/tokenizer_config.json | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2060 >> loading file chat_template.jinja from cache at None | |
| [INFO|2025-04-17 02:09:45] tokenization_utils_base.py:2323 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. | |
| [INFO|2025-04-17 02:09:45] logging.py:143 >> Add pad token: <|eot_id|> | |
| [INFO|2025-04-17 02:09:45] logging.py:143 >> Loading dataset GraceC3/STEInstruct... | |
| [INFO|2025-04-17 02:10:07] configuration_utils.py:693 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/config.json | |
| [INFO|2025-04-17 02:10:07] configuration_utils.py:765 >> Model config LlamaConfig { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 8, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "factor": 32.0, | |
| "high_freq_factor": 4.0, | |
| "low_freq_factor": 1.0, | |
| "original_max_position_embeddings": 8192, | |
| "rope_type": "llama3" | |
| }, | |
| "rope_theta": 500000.0, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.51.1", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } | |
| [INFO|2025-04-17 02:10:07] logging.py:143 >> KV cache is disabled during training. | |
| [INFO|2025-04-17 02:10:08] modeling_utils.py:1124 >> loading weights file model.safetensors from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/model.safetensors.index.json | |
| [INFO|2025-04-17 02:10:52] modeling_utils.py:2167 >> Instantiating LlamaForCausalLM model under default dtype torch.bfloat16. | |
| [INFO|2025-04-17 02:10:52] configuration_utils.py:1142 >> Generate config GenerationConfig { | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "use_cache": false | |
| } | |
| [INFO|2025-04-17 02:11:26] modeling_utils.py:4930 >> All model checkpoint weights were used when initializing LlamaForCausalLM. | |
| [INFO|2025-04-17 02:11:26] modeling_utils.py:4938 >> All the weights of LlamaForCausalLM were initialized from the model checkpoint at meta-llama/Llama-3.2-3B-Instruct. | |
| If your task is similar to the task the model of the checkpoint was trained on, you can already use LlamaForCausalLM for predictions without further training. | |
| [INFO|2025-04-17 02:11:26] configuration_utils.py:1097 >> loading configuration file generation_config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/generation_config.json | |
| [INFO|2025-04-17 02:11:26] configuration_utils.py:1142 >> Generate config GenerationConfig { | |
| "bos_token_id": 128000, | |
| "do_sample": true, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "temperature": 0.6, | |
| "top_p": 0.9 | |
| } | |
| [INFO|2025-04-17 02:11:26] logging.py:143 >> Gradient checkpointing enabled. | |
| [INFO|2025-04-17 02:11:26] logging.py:143 >> Using torch SDPA for faster training and inference. | |
| [INFO|2025-04-17 02:11:26] logging.py:143 >> Upcasting trainable params to float32. | |
| [INFO|2025-04-17 02:11:26] logging.py:143 >> Fine-tuning method: LoRA | |
| [INFO|2025-04-17 02:11:26] logging.py:143 >> Found linear modules: v_proj,down_proj,gate_proj,o_proj,k_proj,q_proj,up_proj | |
| [INFO|2025-04-17 02:11:27] logging.py:143 >> trainable params: 12,156,928 || all params: 3,224,906,752 || trainable%: 0.3770 | |
| [INFO|2025-04-17 02:11:27] trainer.py:748 >> Using auto half precision backend | |
| [WARNING|2025-04-17 02:11:27] trainer.py:783 >> No label_names provided for model class `PeftModelForCausalLM`. Since `PeftModel` hides base models input arguments, if label_names is not given, label_names can't be set automatically within `Trainer`. Note that empty label_names list will be used instead. | |
| [INFO|2025-04-17 02:11:28] trainer.py:2414 >> ***** Running training ***** | |
| [INFO|2025-04-17 02:11:28] trainer.py:2415 >> Num examples = 1,308 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2416 >> Num Epochs = 1 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2417 >> Instantaneous batch size per device = 16 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2420 >> Total train batch size (w. parallel, distributed & accumulation) = 128 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2421 >> Gradient Accumulation steps = 8 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2422 >> Total optimization steps = 10 | |
| [INFO|2025-04-17 02:11:28] trainer.py:2423 >> Number of trainable parameters = 12,156,928 | |
| [INFO|2025-04-17 02:18:51] logging.py:143 >> {'loss': 1.5609, 'learning_rate': 1.3090e-04, 'epoch': 0.49, 'throughput': 112.10} | |
| [INFO|2025-04-17 02:26:24] logging.py:143 >> {'loss': 0.7089, 'learning_rate': 4.8943e-06, 'epoch': 0.98, 'throughput': 111.01} | |
| [INFO|2025-04-17 02:26:24] trainer.py:3984 >> Saving model checkpoint to saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38/checkpoint-10 | |
| [INFO|2025-04-17 02:26:24] configuration_utils.py:693 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/config.json | |
| [INFO|2025-04-17 02:26:24] configuration_utils.py:765 >> Model config LlamaConfig { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 8, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "factor": 32.0, | |
| "high_freq_factor": 4.0, | |
| "low_freq_factor": 1.0, | |
| "original_max_position_embeddings": 8192, | |
| "rope_type": "llama3" | |
| }, | |
| "rope_theta": 500000.0, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.51.1", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } | |
| [INFO|2025-04-17 02:26:25] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38/checkpoint-10/tokenizer_config.json | |
| [INFO|2025-04-17 02:26:25] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38/checkpoint-10/special_tokens_map.json | |
| [INFO|2025-04-17 02:26:25] trainer.py:2681 >> | |
| Training completed. Do not forget to share your model on huggingface.co/models =) | |
| [INFO|2025-04-17 02:26:25] trainer.py:3984 >> Saving model checkpoint to saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38 | |
| [INFO|2025-04-17 02:26:25] configuration_utils.py:693 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/config.json | |
| [INFO|2025-04-17 02:26:25] configuration_utils.py:765 >> Model config LlamaConfig { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 8, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "factor": 32.0, | |
| "high_freq_factor": 4.0, | |
| "low_freq_factor": 1.0, | |
| "original_max_position_embeddings": 8192, | |
| "rope_type": "llama3" | |
| }, | |
| "rope_theta": 500000.0, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.51.1", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } | |
| [INFO|2025-04-17 02:26:26] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38/tokenizer_config.json | |
| [INFO|2025-04-17 02:26:26] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Llama-3.2-3B-Instruct/lora/train_2025-04-17-02-04-38/special_tokens_map.json | |
| [WARNING|2025-04-17 02:26:26] logging.py:148 >> No metric eval_loss to plot. | |
| [WARNING|2025-04-17 02:26:26] logging.py:148 >> No metric eval_accuracy to plot. | |
| [INFO|2025-04-17 02:26:26] modelcard.py:450 >> Dropping the following result as it does not have all the necessary fields: | |
| {'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}} | |