File size: 16,509 Bytes
bdd5ef6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 |
05/30/2024 09:42:38 - INFO - transformers.tokenization_utils_base - loading file tokenizer.json from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/tokenizer.json
05/30/2024 09:42:38 - INFO - transformers.tokenization_utils_base - loading file added_tokens.json from cache at None
05/30/2024 09:42:38 - INFO - transformers.tokenization_utils_base - loading file special_tokens_map.json from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/special_tokens_map.json
05/30/2024 09:42:38 - INFO - transformers.tokenization_utils_base - loading file tokenizer_config.json from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/tokenizer_config.json
05/30/2024 09:42:38 - WARNING - transformers.tokenization_utils_base - Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
05/30/2024 09:42:38 - INFO - llamafactory.data.template - Replace eos token: <|eot_id|>
05/30/2024 09:42:38 - INFO - llamafactory.data.loader - Loading dataset Central-full.json...
05/30/2024 09:42:39 - INFO - transformers.configuration_utils - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/config.json
05/30/2024 09:42:39 - INFO - transformers.configuration_utils - Model config LlamaConfig {
"_name_or_path": "shenzhi-wang/Llama3-8B-Chinese-Chat",
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 128000,
"eos_token_id": 128009,
"hidden_act": "silu",
"hidden_size": 4096,
"initializer_range": 0.02,
"intermediate_size": 14336,
"max_position_embeddings": 8192,
"mlp_bias": false,
"model_type": "llama",
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 8,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 500000.0,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.41.1",
"use_cache": true,
"vocab_size": 128256
}
05/30/2024 09:42:39 - INFO - transformers.modeling_utils - loading weights file model.safetensors from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/model.safetensors.index.json
05/30/2024 09:42:39 - INFO - transformers.modeling_utils - Instantiating LlamaForCausalLM model under default dtype torch.float16.
05/30/2024 09:42:39 - INFO - transformers.generation.configuration_utils - Generate config GenerationConfig {
"bos_token_id": 128000,
"eos_token_id": 128009
}
05/30/2024 09:42:49 - INFO - transformers.modeling_utils - All model checkpoint weights were used when initializing LlamaForCausalLM.
05/30/2024 09:42:49 - INFO - transformers.modeling_utils - All the weights of LlamaForCausalLM were initialized from the model checkpoint at shenzhi-wang/Llama3-8B-Chinese-Chat.
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.
05/30/2024 09:42:49 - INFO - transformers.generation.configuration_utils - loading configuration file generation_config.json from cache at /root/.cache/huggingface/hub/models--shenzhi-wang--Llama3-8B-Chinese-Chat/snapshots/4754413429ccde4f441fe30e44ee62fd1c93b8be/generation_config.json
05/30/2024 09:42:49 - INFO - transformers.generation.configuration_utils - Generate config GenerationConfig {
"bos_token_id": 128000,
"eos_token_id": 128009,
"pad_token_id": 128009
}
05/30/2024 09:42:49 - INFO - llamafactory.model.utils.checkpointing - Gradient checkpointing enabled.
05/30/2024 09:42:49 - INFO - llamafactory.model.utils.attention - Using torch SDPA for faster training and inference.
05/30/2024 09:42:49 - INFO - llamafactory.model.adapter - Upcasting trainable params to float32.
05/30/2024 09:42:49 - INFO - llamafactory.model.adapter - Fine-tuning method: Freeze
05/30/2024 09:42:49 - INFO - llamafactory.model.adapter - Set trainable layers: 30,31
05/30/2024 09:42:49 - INFO - llamafactory.model.loader - trainable params: 436224000 || all params: 8030261248 || trainable%: 5.4323
05/30/2024 09:42:49 - INFO - transformers.trainer - Using auto half precision backend
05/30/2024 09:42:49 - INFO - transformers.trainer - ***** Running training *****
05/30/2024 09:42:49 - INFO - transformers.trainer - Num examples = 766
05/30/2024 09:42:49 - INFO - transformers.trainer - Num Epochs = 3
05/30/2024 09:42:49 - INFO - transformers.trainer - Instantaneous batch size per device = 1
05/30/2024 09:42:49 - INFO - transformers.trainer - Total train batch size (w. parallel, distributed & accumulation) = 8
05/30/2024 09:42:49 - INFO - transformers.trainer - Gradient Accumulation steps = 8
05/30/2024 09:42:49 - INFO - transformers.trainer - Total optimization steps = 285
05/30/2024 09:42:49 - INFO - transformers.trainer - Number of trainable parameters = 436,224,000
05/30/2024 09:42:56 - INFO - llamafactory.extras.callbacks - {'loss': 3.0294, 'learning_rate': 4.9962e-05, 'epoch': 0.05}
05/30/2024 09:43:02 - INFO - llamafactory.extras.callbacks - {'loss': 2.7312, 'learning_rate': 4.9848e-05, 'epoch': 0.10}
05/30/2024 09:43:07 - INFO - llamafactory.extras.callbacks - {'loss': 2.6282, 'learning_rate': 4.9659e-05, 'epoch': 0.16}
05/30/2024 09:43:13 - INFO - llamafactory.extras.callbacks - {'loss': 2.5533, 'learning_rate': 4.9395e-05, 'epoch': 0.21}
05/30/2024 09:43:19 - INFO - llamafactory.extras.callbacks - {'loss': 2.5412, 'learning_rate': 4.9057e-05, 'epoch': 0.26}
05/30/2024 09:43:25 - INFO - llamafactory.extras.callbacks - {'loss': 2.5643, 'learning_rate': 4.8645e-05, 'epoch': 0.31}
05/30/2024 09:43:30 - INFO - llamafactory.extras.callbacks - {'loss': 2.5158, 'learning_rate': 4.8162e-05, 'epoch': 0.37}
05/30/2024 09:43:36 - INFO - llamafactory.extras.callbacks - {'loss': 2.5183, 'learning_rate': 4.7609e-05, 'epoch': 0.42}
05/30/2024 09:43:42 - INFO - llamafactory.extras.callbacks - {'loss': 2.4960, 'learning_rate': 4.6987e-05, 'epoch': 0.47}
05/30/2024 09:43:48 - INFO - llamafactory.extras.callbacks - {'loss': 2.5069, 'learning_rate': 4.6298e-05, 'epoch': 0.52}
05/30/2024 09:43:53 - INFO - llamafactory.extras.callbacks - {'loss': 2.4605, 'learning_rate': 4.5544e-05, 'epoch': 0.57}
05/30/2024 09:43:59 - INFO - llamafactory.extras.callbacks - {'loss': 2.4223, 'learning_rate': 4.4729e-05, 'epoch': 0.63}
05/30/2024 09:44:05 - INFO - llamafactory.extras.callbacks - {'loss': 2.4468, 'learning_rate': 4.3853e-05, 'epoch': 0.68}
05/30/2024 09:44:11 - INFO - llamafactory.extras.callbacks - {'loss': 2.3933, 'learning_rate': 4.2920e-05, 'epoch': 0.73}
05/30/2024 09:44:16 - INFO - llamafactory.extras.callbacks - {'loss': 2.4540, 'learning_rate': 4.1932e-05, 'epoch': 0.78}
05/30/2024 09:44:22 - INFO - llamafactory.extras.callbacks - {'loss': 2.4139, 'learning_rate': 4.0893e-05, 'epoch': 0.84}
05/30/2024 09:44:28 - INFO - llamafactory.extras.callbacks - {'loss': 2.3528, 'learning_rate': 3.9806e-05, 'epoch': 0.89}
05/30/2024 09:44:34 - INFO - llamafactory.extras.callbacks - {'loss': 2.3643, 'learning_rate': 3.8674e-05, 'epoch': 0.94}
05/30/2024 09:44:39 - INFO - llamafactory.extras.callbacks - {'loss': 2.3584, 'learning_rate': 3.7500e-05, 'epoch': 0.99}
05/30/2024 09:44:45 - INFO - llamafactory.extras.callbacks - {'loss': 2.1535, 'learning_rate': 3.6288e-05, 'epoch': 1.04}
05/30/2024 09:44:45 - INFO - transformers.trainer - Saving model checkpoint to saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100
05/30/2024 09:44:45 - INFO - transformers.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100/config.json
05/30/2024 09:44:45 - INFO - transformers.generation.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100/generation_config.json
05/30/2024 09:45:44 - INFO - transformers.modeling_utils - The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 4 checkpoint shards. You can find where each parameters has been saved in the index located at saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100/model.safetensors.index.json.
05/30/2024 09:45:44 - INFO - transformers.tokenization_utils_base - tokenizer config file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100/tokenizer_config.json
05/30/2024 09:45:44 - INFO - transformers.tokenization_utils_base - Special tokens file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-100/special_tokens_map.json
05/30/2024 09:46:01 - INFO - llamafactory.extras.callbacks - {'loss': 2.0786, 'learning_rate': 3.5042e-05, 'epoch': 1.10}
05/30/2024 09:46:07 - INFO - llamafactory.extras.callbacks - {'loss': 2.0251, 'learning_rate': 3.3766e-05, 'epoch': 1.15}
05/30/2024 09:46:12 - INFO - llamafactory.extras.callbacks - {'loss': 2.0486, 'learning_rate': 3.2463e-05, 'epoch': 1.20}
05/30/2024 09:46:18 - INFO - llamafactory.extras.callbacks - {'loss': 2.0030, 'learning_rate': 3.1137e-05, 'epoch': 1.25}
05/30/2024 09:46:24 - INFO - llamafactory.extras.callbacks - {'loss': 2.0196, 'learning_rate': 2.9793e-05, 'epoch': 1.31}
05/30/2024 09:46:30 - INFO - llamafactory.extras.callbacks - {'loss': 1.9855, 'learning_rate': 2.8434e-05, 'epoch': 1.36}
05/30/2024 09:46:36 - INFO - llamafactory.extras.callbacks - {'loss': 2.0136, 'learning_rate': 2.7064e-05, 'epoch': 1.41}
05/30/2024 09:46:41 - INFO - llamafactory.extras.callbacks - {'loss': 1.9636, 'learning_rate': 2.5689e-05, 'epoch': 1.46}
05/30/2024 09:46:47 - INFO - llamafactory.extras.callbacks - {'loss': 1.9941, 'learning_rate': 2.4311e-05, 'epoch': 1.51}
05/30/2024 09:46:53 - INFO - llamafactory.extras.callbacks - {'loss': 1.9606, 'learning_rate': 2.2936e-05, 'epoch': 1.57}
05/30/2024 09:46:59 - INFO - llamafactory.extras.callbacks - {'loss': 2.0351, 'learning_rate': 2.1566e-05, 'epoch': 1.62}
05/30/2024 09:47:04 - INFO - llamafactory.extras.callbacks - {'loss': 1.9508, 'learning_rate': 2.0207e-05, 'epoch': 1.67}
05/30/2024 09:47:10 - INFO - llamafactory.extras.callbacks - {'loss': 1.9504, 'learning_rate': 1.8863e-05, 'epoch': 1.72}
05/30/2024 09:47:16 - INFO - llamafactory.extras.callbacks - {'loss': 1.9508, 'learning_rate': 1.7537e-05, 'epoch': 1.78}
05/30/2024 09:47:22 - INFO - llamafactory.extras.callbacks - {'loss': 1.8806, 'learning_rate': 1.6234e-05, 'epoch': 1.83}
05/30/2024 09:47:28 - INFO - llamafactory.extras.callbacks - {'loss': 1.9759, 'learning_rate': 1.4958e-05, 'epoch': 1.88}
05/30/2024 09:47:33 - INFO - llamafactory.extras.callbacks - {'loss': 1.9918, 'learning_rate': 1.3712e-05, 'epoch': 1.93}
05/30/2024 09:47:39 - INFO - llamafactory.extras.callbacks - {'loss': 1.8922, 'learning_rate': 1.2500e-05, 'epoch': 1.98}
05/30/2024 09:47:45 - INFO - llamafactory.extras.callbacks - {'loss': 1.7482, 'learning_rate': 1.1326e-05, 'epoch': 2.04}
05/30/2024 09:47:51 - INFO - llamafactory.extras.callbacks - {'loss': 1.6307, 'learning_rate': 1.0194e-05, 'epoch': 2.09}
05/30/2024 09:47:51 - INFO - transformers.trainer - Saving model checkpoint to saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200
05/30/2024 09:47:51 - INFO - transformers.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200/config.json
05/30/2024 09:47:51 - INFO - transformers.generation.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200/generation_config.json
05/30/2024 09:48:50 - INFO - transformers.modeling_utils - The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 4 checkpoint shards. You can find where each parameters has been saved in the index located at saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200/model.safetensors.index.json.
05/30/2024 09:48:50 - INFO - transformers.tokenization_utils_base - tokenizer config file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200/tokenizer_config.json
05/30/2024 09:48:50 - INFO - transformers.tokenization_utils_base - Special tokens file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/checkpoint-200/special_tokens_map.json
05/30/2024 09:49:04 - INFO - llamafactory.extras.callbacks - {'loss': 1.6288, 'learning_rate': 9.1069e-06, 'epoch': 2.14}
05/30/2024 09:49:10 - INFO - llamafactory.extras.callbacks - {'loss': 1.6239, 'learning_rate': 8.0680e-06, 'epoch': 2.19}
05/30/2024 09:49:16 - INFO - llamafactory.extras.callbacks - {'loss': 1.6385, 'learning_rate': 7.0804e-06, 'epoch': 2.25}
05/30/2024 09:49:22 - INFO - llamafactory.extras.callbacks - {'loss': 1.6511, 'learning_rate': 6.1473e-06, 'epoch': 2.30}
05/30/2024 09:49:27 - INFO - llamafactory.extras.callbacks - {'loss': 1.6356, 'learning_rate': 5.2715e-06, 'epoch': 2.35}
05/30/2024 09:49:33 - INFO - llamafactory.extras.callbacks - {'loss': 1.6163, 'learning_rate': 4.4556e-06, 'epoch': 2.40}
05/30/2024 09:49:39 - INFO - llamafactory.extras.callbacks - {'loss': 1.6386, 'learning_rate': 3.7020e-06, 'epoch': 2.45}
05/30/2024 09:49:45 - INFO - llamafactory.extras.callbacks - {'loss': 1.6722, 'learning_rate': 3.0132e-06, 'epoch': 2.51}
05/30/2024 09:49:50 - INFO - llamafactory.extras.callbacks - {'loss': 1.6134, 'learning_rate': 2.3911e-06, 'epoch': 2.56}
05/30/2024 09:49:56 - INFO - llamafactory.extras.callbacks - {'loss': 1.6621, 'learning_rate': 1.8376e-06, 'epoch': 2.61}
05/30/2024 09:50:02 - INFO - llamafactory.extras.callbacks - {'loss': 1.5881, 'learning_rate': 1.3546e-06, 'epoch': 2.66}
05/30/2024 09:50:08 - INFO - llamafactory.extras.callbacks - {'loss': 1.6241, 'learning_rate': 9.4330e-07, 'epoch': 2.72}
05/30/2024 09:50:14 - INFO - llamafactory.extras.callbacks - {'loss': 1.5779, 'learning_rate': 6.0509e-07, 'epoch': 2.77}
05/30/2024 09:50:19 - INFO - llamafactory.extras.callbacks - {'loss': 1.5785, 'learning_rate': 3.4097e-07, 'epoch': 2.82}
05/30/2024 09:50:25 - INFO - llamafactory.extras.callbacks - {'loss': 1.6248, 'learning_rate': 1.5173e-07, 'epoch': 2.87}
05/30/2024 09:50:31 - INFO - llamafactory.extras.callbacks - {'loss': 1.5898, 'learning_rate': 3.7962e-08, 'epoch': 2.92}
05/30/2024 09:50:37 - INFO - llamafactory.extras.callbacks - {'loss': 1.5727, 'learning_rate': 0.0000e+00, 'epoch': 2.98}
05/30/2024 09:50:37 - INFO - transformers.trainer -
Training completed. Do not forget to share your model on huggingface.co/models =)
05/30/2024 09:50:37 - INFO - transformers.trainer - Saving model checkpoint to saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42
05/30/2024 09:50:37 - INFO - transformers.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/config.json
05/30/2024 09:50:37 - INFO - transformers.generation.configuration_utils - Configuration saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/generation_config.json
05/30/2024 09:51:37 - INFO - transformers.modeling_utils - The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 4 checkpoint shards. You can find where each parameters has been saved in the index located at saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/model.safetensors.index.json.
05/30/2024 09:51:37 - INFO - transformers.tokenization_utils_base - tokenizer config file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/tokenizer_config.json
05/30/2024 09:51:37 - INFO - transformers.tokenization_utils_base - Special tokens file saved in saves/LLaMA3-8B-Chinese-Chat/freeze/train_2024-05-30-09-37-42/special_tokens_map.json
05/30/2024 09:51:38 - WARNING - llamafactory.extras.ploting - No metric eval_loss to plot.
05/30/2024 09:51:38 - INFO - transformers.modelcard - Dropping the following result as it does not have all the necessary fields:
{'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}}
|