[INFO|2025-03-07 10:08:21] configuration_utils.py:771 >> Model config LlamaConfig {
  "_name_or_path": "meta-llama/Llama-3.3-70B-Instruct",
  "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": 8192,
  "initializer_range": 0.02,
  "intermediate_size": 28672,
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 64,
  "num_hidden_layers": 80,
  "num_key_value_heads": 8,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "factor": 8.0,
    "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": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.49.0",
  "use_cache": true,
  "vocab_size": 128256
}


[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file tokenizer.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/tokenizer.json

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file tokenizer.model from cache at None

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file added_tokens.json from cache at None

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file special_tokens_map.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/special_tokens_map.json

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file tokenizer_config.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/tokenizer_config.json

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2050 >> loading file chat_template.jinja from cache at None

[INFO|2025-03-07 10:08:21] tokenization_utils_base.py:2313 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.

[INFO|2025-03-07 10:08:22] configuration_utils.py:699 >> loading configuration file config.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/config.json

[INFO|2025-03-07 10:08:22] configuration_utils.py:771 >> Model config LlamaConfig {
  "_name_or_path": "meta-llama/Llama-3.3-70B-Instruct",
  "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": 8192,
  "initializer_range": 0.02,
  "intermediate_size": 28672,
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 64,
  "num_hidden_layers": 80,
  "num_key_value_heads": 8,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "factor": 8.0,
    "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": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.49.0",
  "use_cache": true,
  "vocab_size": 128256
}


[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file tokenizer.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/tokenizer.json

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file tokenizer.model from cache at None

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file added_tokens.json from cache at None

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file special_tokens_map.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/special_tokens_map.json

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file tokenizer_config.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/tokenizer_config.json

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2050 >> loading file chat_template.jinja from cache at None

[INFO|2025-03-07 10:08:22] tokenization_utils_base.py:2313 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.

[INFO|2025-03-07 10:08:22] logging.py:157 >> Add <|eot_id|>,<|eom_id|> to stop words.

[INFO|2025-03-07 10:08:22] logging.py:157 >> Loading dataset jgayed/ets480...

[INFO|2025-03-07 10:08:27] configuration_utils.py:699 >> loading configuration file config.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/config.json

[INFO|2025-03-07 10:08:27] configuration_utils.py:771 >> Model config LlamaConfig {
  "_name_or_path": "meta-llama/Llama-3.3-70B-Instruct",
  "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": 8192,
  "initializer_range": 0.02,
  "intermediate_size": 28672,
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 64,
  "num_hidden_layers": 80,
  "num_key_value_heads": 8,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "factor": 8.0,
    "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": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.49.0",
  "use_cache": true,
  "vocab_size": 128256
}


[INFO|2025-03-07 10:08:27] logging.py:157 >> Quantizing model to 4 bit with bitsandbytes.

[INFO|2025-03-07 10:08:27] quantizer_bnb_4bit.py:276 >> The device_map was not initialized. Setting device_map to {'': 0}. If you want to use the model for inference, please set device_map ='auto' 

[INFO|2025-03-07 10:08:27] modeling_utils.py:3982 >> loading weights file model.safetensors from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/model.safetensors.index.json

[INFO|2025-03-07 10:08:27] modeling_utils.py:1633 >> Instantiating LlamaForCausalLM model under default dtype torch.bfloat16.

[INFO|2025-03-07 10:08:27] configuration_utils.py:1140 >> Generate config GenerationConfig {
  "bos_token_id": 128000,
  "eos_token_id": [
    128001,
    128008,
    128009
  ]
}


[INFO|2025-03-07 10:09:51] modeling_utils.py:4970 >> All model checkpoint weights were used when initializing LlamaForCausalLM.


[INFO|2025-03-07 10:09:51] modeling_utils.py:4978 >> All the weights of LlamaForCausalLM were initialized from the model checkpoint at meta-llama/Llama-3.3-70B-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|2025-03-07 10:09:51] configuration_utils.py:1095 >> loading configuration file generation_config.json from cache at /workspace/hf_home/hub/models--meta-llama--Llama-3.3-70B-Instruct/snapshots/6f6073b423013f6a7d4d9f39144961bfbfbc386b/generation_config.json

[INFO|2025-03-07 10:09:51] configuration_utils.py:1140 >> Generate config GenerationConfig {
  "bos_token_id": 128000,
  "do_sample": true,
  "eos_token_id": [
    128001,
    128008,
    128009
  ],
  "temperature": 0.6,
  "top_p": 0.9
}


[INFO|2025-03-07 10:09:51] logging.py:157 >> Gradient checkpointing enabled.

[INFO|2025-03-07 10:09:51] logging.py:157 >> Using torch SDPA for faster training and inference.

[INFO|2025-03-07 10:09:51] logging.py:157 >> Upcasting trainable params to float32.

[INFO|2025-03-07 10:09:51] logging.py:157 >> Fine-tuning method: LoRA

[INFO|2025-03-07 10:09:51] logging.py:157 >> Found linear modules: up_proj,o_proj,v_proj,q_proj,k_proj,gate_proj,down_proj

[WARNING|2025-03-07 10:09:54] trainer.py:781 >> 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|2025-03-07 10:09:56] logging.py:157 >> trainable params: 828,375,040 || all params: 276,735,205,376 || trainable%: 0.2993

[INFO|2025-03-07 10:09:56] trainer.py:746 >> Using auto half precision backend

[WARNING|2025-03-07 10:09:56] trainer.py:781 >> 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|2025-03-07 10:09:56] deepspeed.py:334 >> Detected ZeRO Offload and non-DeepSpeed optimizers: This combination should work as long as the custom optimizer has both CPU and GPU implementation (except LAMB)