CodeGen-HTML 350M Open-Source Model - A Practical Tool for Free High-Quality HTML Code Generation

Codegen 350m Html

Developed by alecsharpie

CodeGen-HTML 350M is an autoregressive language model fine-tuned from CodeGen-Multi 350M, specifically designed for HTML code generation.

Large Language Model

Transformers

OtherOpen Source License:Bsd-3-clause #HTML code generation #Multilingual pre-training #Autoregressive model

Downloads 132

Release Time : 12/1/2022

Model Overview

This model is an autoregressive language model for program synthesis, optimized particularly for HTML code generation. It can generate complete HTML code based on natural language prompts or partial code snippets.

Model Features

HTML code generation

Fine-tuned specifically for HTML code generation, capable of producing high-quality HTML code snippets.

Autoregressive model

Utilizes an autoregressive architecture to predict and generate subsequent code based on context.

Multilingual foundation

Based on the CodeGen-Multi 350M model, with comprehension capabilities for multiple programming languages.

Model Capabilities

HTML code generation

Code completion

Program synthesis

Use Cases

Web development

HTML template generation

Generates complete HTML page structures based on simple descriptions.

Produces standards-compliant HTML code

Code completion

Automatically completes remaining HTML code based on partial snippets.

Improves development efficiency

🚀 CodeGen (CodeGen-HTML 350M)

CodeGen is an autoregressive language model family for program synthesis, offering diverse pre - training and size options.

🚀 Quick Start

This model can be easily loaded using the AutoModelForCausalLM functionality:

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Salesforce/codegen-350M-multi")
model = AutoModelForCausalLM.from_pretrained("alecsharpie/codegen_350m_html")

text = "<body>"

input_ids = tokenizer(text, return_tensors="pt").input_ids
generated_ids = model.generate(input_ids, max_length=128)
print(tokenizer.decode(generated_ids[0], skip_special_tokens=True))

✨ Features

Program Synthesis: CodeGen is designed for program synthesis, capable of generating executable code from English prompts in the form of comment strings and completing partially - generated code.
Multiple Variants: Available in 3 pre - training data variants (NL, Multi, Mono) and 4 model size variants (350M, 2B, 6B, 16B).

📚 Documentation

Model description

CodeGen is a family of autoregressive language models for program synthesis from the paper: A Conversational Paradigm for Program Synthesis by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong. The models are originally released in this repository, under 3 pre - training data variants (NL, Multi, Mono) and 4 model size variants (350M, 2B, 6B, 16B).

The checkpoint included in this repository is finetuned on top of the CodeGen - Multi 350M, where "Multi" means the model is initialized with CodeGen - NL 350M and further pre - trained on a dataset of multiple programming languages, and "350M" refers to the number of trainable parameters.

It has been finetuned on HTML code contained in bigcode/the - stack dataset on huggingface.

Training data

This checkpoint (CodeGen - Multi 350M) was firstly initialized with CodeGen - NL 350M, and then pre - trained on [BigQuery](https://console.cloud.google.com/marketplace/details/github/github - repos), a large - scale dataset of multiple programming languages from GitHub repositories. The data consists of 119.2B tokens and includes C, C++, Go, Java, JavaScript, and Python.

Lastly it has been finetuned on HTML code contained in [bigcode/the - stack](https://huggingface.co/datasets/bigcode/the - stack) dataset on huggingface

Training procedure

Initially: CodeGen was trained using cross - entropy loss to maximize the likelihood of sequential inputs. The family of models are trained using multiple TPU - v4 - 512 by Google, leveraging data and model parallelism. See Section 2.3 of the paper for more details.

Finetune: I fine tuned the 350M model on a single A100 with 40Gb of RAM, with batch size 10 and an input length of 512 tokens Used 80 - 90% of the RAM

Intended Use and Limitations

As an autoregressive language model, CodeGen is capable of extracting features from given natural language and programming language texts, and calculating the likelihood of them. However, the model is intended for and best at program synthesis, that is, generating executable code given English prompts, where the prompts should be in the form of a comment string. The model can complete partially - generated code as well.

🔧 Technical Details

Training Loss: CodeGen was trained using cross - entropy loss to maximize the likelihood of sequential inputs.
Hardware Utilization: The initial training used multiple TPU - v4 - 512 by Google with data and model parallelism. The finetuning of the 350M model was done on a single A100 with 40Gb of RAM, batch size 10, and an input length of 512 tokens, using 80 - 90% of the RAM.

📄 License

This project is licensed under the BSD 3 - Clause license.

📚 Citation

@article{Nijkamp2022ACP,
  title={A Conversational Paradigm for Program Synthesis},
  author={Nijkamp, Erik and Pang, Bo and Hayashi, Hiroaki and Tu, Lifu and Wang, Huan and Zhou, Yingbo and Savarese, Silvio and Xiong, Caiming},
  journal={arXiv preprint},
  year={2022}
}

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご