{"id":250,"date":"2023-09-08T20:58:20","date_gmt":"2023-09-08T20:58:20","guid":{"rendered":"https:\/\/techwyns.com\/tech\/?p=250"},"modified":"2023-09-08T21:01:34","modified_gmt":"2023-09-08T21:01:34","slug":"artificial-intelligence-opensource-alternatives-to-chatgpt","status":"publish","type":"post","link":"https:\/\/techwyns.com\/tech\/blog\/2023\/09\/08\/artificial-intelligence-opensource-alternatives-to-chatgpt\/","title":{"rendered":"Artificial intelligence – OpenSource alternatives to ChatGPT"},"content":{"rendered":"
<\/p>\n
\n
\n
\n
\n
Top 9 Free GPT-3 Alternative AI models<\/h2>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
\n
\n
\n
\n
\n
Now that you\u2019ve got an idea of what is the technology we are talking about, let\u2019s move on to the\u00a0OpenAI GPT-3\u00a0competitors.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/section>\n
\n
\n
\n
\n
OPT by Meta<\/h3>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
\n
\n
\n
\n
\n
Another solid\u00a0GPT-3 open-source\u00a0alternative was released by Meta in May 2022.\u00a0Open Pretrained Transformer language\u00a0model (OPT for short) contains 175B parameters. OPT was trained on multiple public datasets, including The Pile and BookCorpus.<\/p>\n
Its main distinctive feature is that OPT combines both pretrained models and the source code for using or training them.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/section>\n
<\/p>\n
\n
\n
\n
\n
AlexaTM by Amazon<\/h3>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
\n
\n
\n
\n
\n
On November 18, 2022, Amazon publicly released\u00a0AlexaTM\u00a020B, a large-scale multilingual sequence2sequence model. What\u2019s so special about it? It utilizes an\u00a0encoder-decoder architecture\u00a0and was trained on a combo of causal-language-modeling (CLM) and denoising tasks.<\/p>\n
Thanks to that, AlexaTM is a better few-shot learner than decoder-only models. As a result, it\u00a0performs better in 1-shot summarization and machine translation\u00a0tasks than Google\u2019s PaLM 540B. Also, in zero-shot testing, the model tops GPT-3 on SuperGlue and SQuADv2 datasets.<\/p>\n
As for less technical things, AlexaTM supports multiple languages (as its type implies), including English, Spanish, Arabic, German, Hindi, French, Japanese, Italian, Portuguese, and others.<\/p>\n
Altogether, this makes AlexaTM quite a notable opponent to all other LLMs, whether free or not.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/section>\n
\n
\n
\n
\n
GPT-J and GPT-NeoX by EleutherAI<\/h3>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
\n
\n
\n
\n
\n
GPT-J\u00a0is a small 6B-parameter autoregressive model for text generation, completely free to use. It was trained on The Pile, a dataset with 22 subsets of more than 800 GB of English texts.<\/p>\n
Despite its small size, the model performs nearly the same as GPT-3 6.7B-param and is better than its predecessor, GPT-Neo. The latter had 2 versions, 1.3 and 2.7 billion, and in February 2022 grew into\u00a0GPT-NeoX, containing 20B parameters.<\/p>\n
Here\u2019s a quick overview of how GPT-J and GPT-NeoX perform compared to\u00a0OpenAI GPT-3\u00a0versions.<\/p>\n<\/div>\n<\/div>\n
\n
<\/picture><\/div>\n<\/div>\n
\n
\n
As you might see, there\u2019s little to no difference in performance between open-source GPT-J and GPT-NeoX and paid GPT-3 models.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/section>\n
\n
\n
\n
\n
Jurassic-1 language model by AI21 labs<\/h3>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
\n
\n
\n
\n
\n
Jurassic-1 is an autoregressive natural language processing (NLP) model, available in open beta for developers and researchers.<\/p>\n
Yet, it\u2019s not fully open-source, but upon registration, you get $90 credits for free. You can use those credits in the playground with the pre-designed templates for rephrasing, summarization, writing, chatting, drafting outlines, tweeting, coding, and more. What\u2019s more, you can create and train your custom models.<\/p>\n
Jurassic-1\u00a0might become quite a serious rival to GPT-3, as it consists of 2 parts: J1-Jumbo, trained on over 178B parameters, and J1-Large, consisting of 7B parameters. This already makes it 3B parameters more advanced than GPT-3 language model.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/section>\n
\n
\n
\n
\n
CodeGen by Salesforce<\/h3>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n\n
![\"\"](\"https:\/\/www.altamira.ai\/wp-content\/uploads\/2023\/01\/Accuracy-standard-language-modeling-tasks-.png\")
<\/picture><\/div>\n<\/div>\n\n![\"\"](\"https:\/\/www.altamira.ai\/wp-content\/uploads\/2023\/01\/Accuracy-standard-language-modeling-tasks-.png.webp)
<\/picture><\/div>\n<\/div>\n