Picking Up from the Previous Llama Factory Fine-Tuning
First, I plan to use a local dataset for training
Define the training file acer_faqs.json in the local faqs folder
[
{
"instruction": "Explain how much a good gaming PC costs.",
"input": "How much is a good gaming PC?",
"output": "There are many ways to define a good gaming PC, as different gamers have different needs. At Acer, you can purchase a solid, entry-level gaming PC for as little as $849. If you are looking for a high-performance PC that can handle the highest resolutions at high speeds, a good gaming PC would be around $4,000. On average, good gaming PCs range between $1,000 and $4,000."
},
{
"instruction": "List some good gaming PC options.",
"input": "What are good gaming PCs?",
"output": "Many manufacturers and builders offer good gaming PCs with quality components. Acer has a large inventory of great gaming PCs, including both Intel and AMD desktops. The latest 12th generation Intel desktops and Ryzen 9 desktops each provide some of the most seamless gaming experiences."
},
{
"instruction": "Define what a gaming PC is.",
"input": "What is a gaming PC?",
"output": "Gaming PCs are specially designed personal computers made for playing video games with high quality. They are more powerful and have more advanced components than typical computers to provide better graphics and working memory to handle the load of running games."
},
{
"instruction": "Where can someone buy a gaming PC?",
"input": "Where to buy a gaming PC?",
"output": "When considering buying a gaming PC, you have many options between in-store and online retailers. Acer offers great deals and discounts on gaming PCs that use top-tier components."
}
]
Next, place the training data folder into LLaMA-Factory\data\
Modify dataset_info.json to include the file we just defined. The acer_faqs dataset will then appear in the dataset dropdown menu.
{
....
"acer_faqs": {
"file_name": "acer_faqs.json"
}
...
}
Of course, you can also get open-source datasets to train your AI.
You can also use open-source datasets to train your AI model by modifying dataset_info.json.
Starting the Training
Next, let's test the fine-tuned AI model
I tried a few datasets, including the default wiki_faq dataset. It was pretty funny watching the AI I spent half a day training just spout nonsense.
A quick Google search revealed that many people have run into this issue of nonsensical answers. I suspect that the VRAM on a typical PC graphics card is only sufficient for training these stripped-down versions of open-source models. To get more accurate results, you probably need to use the full model and train it on dedicated AI chips.
Final Thoughts
After watching the AI I spent half a day training do nothing but give nonsensical answers, it seems that if the goal is to build a functional AI bot, using pre-trained models and AI Agent technologies is a much more achievable path for now.
To train an AI yourself, you need a massive amount of data. Fine-tuning large open-source models has steep requirements for graphics cards, and there are many complex parameters to learn. I don't think AI training technology is yet easy enough to operate or accessible enough for the average company to readily adopt. It makes you wonder just how much money and effort were poured into building the major AI models we see today.
I'm a complete beginner when it comes to AI training, so if I've misunderstood anything, please feel free to discuss it in the comments or send me a private message.
Comments