Spaces:
Running
Running
| <html> | |
| <head> | |
| <meta charset="utf-8" /> | |
| <meta name="viewport" content="width=device-width" /> | |
| <title>My static Space</title> | |
| <link rel="stylesheet" href="style.css" /> | |
| </head> | |
| <body> | |
| <div class="card"> | |
| <h1>Welcome to your static Space!</h1> | |
| # SuperTransformer | |
| Suptertransformer that auto loads Huggingface models | |
| # Introduction | |
| This is a single line transformer for easy to load models from Huggingface. It is not to replace Huggingface Transformer process. It simplifies it and speed up the loading the process of the HuggingFace models | |
| # Usage | |
| SuperTransformers download the model locally. The super class uses AutoTokenizer and AutoModelForCausalLM.from_pretrained. | |
| # Installation | |
| ``` bash | |
| pip install bitsandbytes>=0.39.0 | |
| pip install --upgrade accelerate transformers | |
| ``` | |
| # How to run | |
| ```python | |
| python SuperTransformer.py | |
| ``` | |
| # Example of usage: | |
| ```python | |
| # Load SuperTransformer Class, (1) Loads Huggingface model, (2) System Prompt (3) Text/prompt (4)Max tokens | |
| SuperTransformers = SuperTransformers("EpistemeAI/ReasoningCore-3B-RE1-V2","You are a highly knowledgeable assistant with expertise in chemistry and physics. <reasoning>","What is the area of a circle, radius=16, reason step by step", 2026) | |
| # 8-bit quantization | |
| SuperTransformers.HuggingFaceTransformer8bit() | |
| # or 4-bit quantization | |
| SuperTransformers.HuggingFaceTransformer4bit() | |
| ``` | |
| ## Returns model and tokenizer | |
| ```python | |
| SuperTransformers = SuperTransformers("EpistemeAI/ReasoningCore-3B-RE1-V2") | |
| model, tokenizer = HuggingfaceTransfomer() #returns the model and tokenizer | |
| ``` | |
| ## returns pipline as higher helper | |
| ```python | |
| SuperTransformers = SuperTransformers("EpistemeAI/ReasoningCore-3B-RE1-V2") | |
| pipe = HuggingfacePipeline() #returns the pipeline only | |
| output = pipe(self.text, max_new_tokens=self.max_new_tokens) # Limit output length to save memory | |
| # Print the generated output | |
| print(output) | |
| </p> | |
| </div> | |
| </body> | |
| </html> | |