Lila-103B-L2 needs about 42.4 GB of memory at Q2_K (a 37.9 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 13 quants, from 21.7 GB (IQ1_S) to 49.6 GB (Q3_K_M).
Lila-103B-L2 is a 103B-parameter Lila model in GGUF format, quantized by mradermacher. Its trained context length is 4K tokens. The repository has 13 quant files, from 21.7 GB (IQ1_S) to 49.6 GB (Q3_K_M).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 21.7 GB | 24.2 GB | Too big | Get file → | |
| IQ1_M | 23.8 GB | 26.3 GB | Too big | Get file → | |
| IQ2_XXS | 27.3 GB | 29.8 GB | Too big | Get file → | |
| IQ2_XS | 30.4 GB | 32.9 GB | Too big | Get file → | |
| IQ2_S | 31.9 GB | 34.4 GB | Too big | Get file → | |
| IQ2_M | 34.7 GB | 37.2 GB | Too big | Get file → | |
| Q2_K | 37.9 GB | 40.4 GB | Too big | Get file → | |
| IQ3_XXS | 39.6 GB | 42.1 GB | Too big | Get file → | |
| IQ3_XS | 42.2 GB | 44.7 GB | Too big | Get file → | |
| Q3_K_S | 44.5 GB | 47.0 GB | Too big | Get file → | |
| IQ3_S | 44.6 GB | 47.1 GB | Too big | Get file → | |
| IQ3_M | 46.1 GB | 48.6 GB | Too big | Get file → | |
| Q3_K_M | 49.6 GB | 52.1 GB | Too big | Get file → |
What GPU do you need for Lila-103B-L2?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: the best quant that fits is IQ2_XXS (27.3 GB file, about 31.8 GB in use).
- 48 GB VRAM: the best quant that fits is IQ3_XS (42.2 GB file, about 46.7 GB in use).
Which Lila-103B-L2 quant should you choose?
Q2_K (37.9 GB) is the middle option in this repository. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
If memory is tight, the smallest file is IQ1_S at 21.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Lila-103B-L2
How much VRAM does Lila-103B-L2 need?
About 42.4 GB at Q2_K with an 8K context window. That is the 37.9 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Lila-103B-L2 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 26.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Lila-103B-L2 GGUF?
It depends on the quant: IQ1_S is 21.7 GB, IQ1_M is 23.8 GB, IQ2_XXS is 27.3 GB, IQ2_XS is 30.4 GB, IQ2_S is 31.9 GB, IQ2_M is 34.7 GB and 7 more.
What context length does Lila-103B-L2 support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 4.0 GB at 8K and 2.0 GB at 4K.
Where can I download Lila-103B-L2 GGUF?
The files are on Hugging Face in the mradermacher/Lila-103B-L2-i1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Lila-103B-L2 use?
Hugging Face lists the license as cc-by-nc-4.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.