LLaMA LLM: Versions, Prompt Templates & Hardware Requirements
Explore all versions of the model, their file formats like GGML, GPTQ, and HF, and understand the hardware requirements for local inference.
Meta has released LLaMA (v1) (Large Language Model Meta AI), a foundational language model designed to assist researchers in the AI field. LLaMA distinguishes itself due to its smaller, more efficient size, making it less resource-intensive than some other large models. These foundation models train on vast amounts of unlabeled data, allowing them to be tailored for a multitude of tasks. LLaMA is available in various sizes, including 7B, 13B, 33B, and 65B parameters.
The model was trained using text from the 20 languages with the highest number of speakers, primarily focusing on those with Latin and Cyrillic scripts. While the model was designed to be versatile for a wide range of applications, challenges like potential biases remain. To manage responsible dissemination, Meta provides LLaMA under a noncommercial license, granting access mainly for research purposes on a case-by-case basis.
Addtional information about LLaMA (v1)
LLaMA (v1) quickly established itself as a foundational model in the AI realm, serving as a versatile platform for numerous fine-tuned variations. Its efficient design, combined with its capacity to train on extensive unlabeled data, made it an ideal base for researchers and developers to build upon. This adaptability catalyzed a surge in merges and fine-tuning exercises, resulting in specialized models tailored to specific tasks and applications.
Furthermore, LLaMA's accessibility in terms of size and resource requirements ignited a trend where more local systems could run Large Language models, democratizing AI capabilities beyond the confines of high-resource environments. This evolution showcased the transformative potential of base models like LLaMA, not only as standalone giants but as foundational tools driving innovation across the AI spectrum.
Notable models that emerged from its robust base include Wizard-Vicuna, Airoboros, WizardLM, Vicuna, and Guanaco. These derived models, each with its unique set of capabilities and focus areas
Hardware requirements
The performance of an LLaMA model depends heavily on the hardware it's running on. For recommendations on the best computer hardware configurations to handle LLaMA models smoothly, check out this guide: Best Computer for Running LLaMA and LLama-2 Models.
Below are the LLaMA hardware requirements for 4-bit quantization:
For 7B Parameter Models
If the 7B llama-13b-supercot-GGML model is what you're after, you gotta think about hardware in two ways. First, for the GPTQ version, you'll want a decent GPU with at least 6GB VRAM. The GTX 1660 or 2060, AMD 5700 XT, or RTX 3050 or 3060 would all work nicely. But for the GGML / GGUF format, it's more about having enough RAM. You'll need around 4 gigs free to run that one smoothly.
| Format | RAM Requirements | VRAM Requirements |
|---|---|---|
| GPTQ (GPU inference) | 6GB (Swap to Load*) | 6GB |
| GGML / GGUF (CPU inference) | 4GB | 300MB |
| Combination of GPTQ and GGML / GGUF (offloading) | 2GB | 2GB |
*RAM needed to load the model initially. Not required for inference. If your system doesn't have quite enough RAM to fully load the model at startup, you can create a swap file to help with the loading.
For 13B Parameter Models
For beefier models like the llama-13b-supercot-GGML, you'll need more powerful hardware. If you're using the GPTQ version, you'll want a strong GPU with at least 10 gigs of VRAM. AMD 6900 XT, RTX 2060 12GB, RTX 3060 12GB, or RTX 3080 would do the trick. For the CPU infgerence (GGML / GGUF) format, having enough RAM is key. You'll want your system to have around 8 gigs available to run it smoothly.
| Format | RAM Requirements | VRAM Requirements |
|---|---|---|
| GPTQ (GPU inference) | 12GB (Swap to Load*) | 10GB |
| GGML / GGUF (CPU inference) | 8GB | 500MB |
| Combination of GPTQ and GGML / GGUF (offloading) | 10GB | 10GB |
*RAM needed to load the model initially. Not required for inference. If your system doesn't have quite enough RAM to fully load the model at startup, you can create a swap file to help with the loading.
For 30B, 33B, and 34B Parameter Models
If you're venturing into the realm of larger models the hardware requirements shift noticeably. GPTQ models benefit from GPUs like the RTX 3080 20GB, A4500, A5000, and the likes, demanding roughly 20GB of VRAM. Conversely, GGML formatted models will require a significant chunk of your system's RAM, nearing 20 GB.
| Format | RAM Requirements | VRAM Requirements |
|---|---|---|
| GPTQ (GPU inference) | 32GB (Swap to Load*) | 20GB |
| GGML / GGUF (CPU inference) | 20GB | 500MB |
| Combination of GPTQ and GGML / GGUF (offloading) | 10GB | 4GB |
*RAM needed to load the model initially. Not required for inference. If your system doesn't have quite enough RAM to fully load the model at startup, you can create a swap file to help with the loading.
For 65B and 70B Parameter Models
When you step up to the big models like 65B and 70B models (llama-65B-GGML), you need some serious hardware. For GPU inference and GPTQ formats, you'll want a top-shelf GPU with at least 40GB of VRAM. We're talking an A100 40GB, dual RTX 3090s or 4090s, A40, RTX A6000, or 8000. You'll also need 64GB of system RAM. For GGML / GGUF CPU inference, have around 40GB of RAM available for both the 65B and 70B models.
| Format | RAM Requirements | VRAM Requirements |
|---|---|---|
| GPTQ (GPU inference) | 64GB (Swap to Load*) | 40GB |
| GGML / GGUF (CPU inference) | 40GB | 600MB |
| Combination of GPTQ and GGML / GGUF (offloading) | 20GB | 20GB |
*RAM needed to load the model initially. Not required for inference. If your system doesn't have quite enough RAM to fully load the model at startup, you can create a swap file to help with the loading.
Memory speed
When running LLaMA AI models, you gotta pay attention to how RAM bandwidth and mdodel size impact inference speed. These large language models need to load completely into RAM or VRAM each time they generate a new token (piece of text). For example, a 4-bit 7B billion parameter LLaMA model takes up around 4.0GB of RAM.
Suppose your have Ryzen 5 5600X processor and DDR4-3200 RAM with theoretical max bandwidth of 50 GBps. In this scenario, you can expect to generate approximately 9 tokens per second. Typically, this performance is about 70% of your theoretical maximum speed due to several limiting factors such as inference sofware, latency, system overhead, and workload characteristics, which prevent reaching the peak speed. To achieve a higher inference speed, say 16 tokens per second, you would need more bandwidth. For example, a system with DDR5-5600 offering around 90 GBps could be enough.
For comparison, high-end GPUs like the Nvidia RTX 3090 boast nearly 930 GBps of bandwidth for their VRAM. The DDR5-6400 RAM can provide up to 100 GB/s. Therefore, understanding and optimizing bandwidth is crucial for running models like LLaMA efficiently
Recommendations:
- For Best Performance: Opt for a machine with a high-end GPU (like NVIDIA's latest RTX 3090 or RTX 4090) or dual GPU setup to accommodate the largest models (65B and 70B). A system with adequate RAM (minimum 16 GB, but 64 GB best) would be optimal.
- For Budget Constraints: If you're limited by budget, focus on LLaMA GGML/GGUF models that fit within the sytem RAM. Remember, while you can offload some weights to the system RAM, it will come at a performance cost.
Remember, these are recommendations, and the actual performance will depend on several factors, including the specific task, model implementation, and other system processes.
CPU requirements
For best performance, a modern multi-core CPU is recommended. An Intel Core i7 from 8th gen onward or AMD Ryzen 5 from 3rd gen onward will work well. CPU with 6-core or 8-core is ideal. Higher clock speeds also improve prompt processing, so aim for 3.6GHz or more.
Having CPU instruction sets like AVX, AVX2, AVX-512 can further improve performance if available. The key is to have a reasonably modern consumer-level CPU with decent core count and clocks, along with baseline vector processing (required for CPU inference with llama.cpp) through AVX2. With those specs, the CPU should handle LLaMA model size.