From b3bf9d7795505eaa1817c00e2dfdf304d60adb7c Mon Sep 17 00:00:00 2001 From: coffee Date: Tue, 29 Apr 2025 19:05:11 +0800 Subject: [PATCH] update LLM --- test.py | 3 --- test_huggingface_train.py | 0 test_huggingface_transformer.py | 18 ------------------ 3 files changed, 21 deletions(-) delete mode 100644 test.py delete mode 100644 test_huggingface_train.py delete mode 100644 test_huggingface_transformer.py diff --git a/test.py b/test.py deleted file mode 100644 index 11acf7e..0000000 --- a/test.py +++ /dev/null @@ -1,3 +0,0 @@ -from transformers import AutoTokenizer -tokenizer = AutoTokenizer.from_pretrained("tokenizer_file/hf_tokenizer") -print(tokenizer.vocab_size) diff --git a/test_huggingface_train.py b/test_huggingface_train.py deleted file mode 100644 index e69de29..0000000 diff --git a/test_huggingface_transformer.py b/test_huggingface_transformer.py deleted file mode 100644 index c4e6ed9..0000000 --- a/test_huggingface_transformer.py +++ /dev/null @@ -1,18 +0,0 @@ -import transformers -from transformers import AutoModelForCausalLM, AutoTokenizer -from transformers import LlamaForCausalLM -tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-0.5B-Instruct") -from transformers import LlamaConfig -config = LlamaConfig( - vocab_size=32000, - hidden_size=4096, - num_attention_heads=32, - num_hidden_layers=32, - intermediate_size=11008, -) - -model = LlamaForCausalLM(config) - -model = transformers.Qwen2ForCausalLM(config) -model_size = sum(t.numel() for t in model.parameters()) -print(f"Model Size: {model_size/1000**2:.1f}M parameters") \ No newline at end of file