LI

LilMistral DPO 300M

LilmistralGGUF by mradermacherUpdated 2024-08-01
Download on Hugging Face
Parameters300M
Context4K
Layers20
Architecturellama
Smallest quant0.2 GB
Quick answer

LilMistral DPO 300M needs about 1.0 GB of memory at Q4_K_M (a 0.2 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 15 quants, from 0.2 GB (IQ3_S) to 0.6 GB (F16).

LilMistral DPO 300M – a lightweight 300M‑parameter Llama model

LilMistral DPO 300M is a 300M‑parameter Llama architecture model trained with a 4096 token context. It is released under the apache-2.0 license and available in multiple GGUF quantizations. The model is designed for efficient inference on modest hardware.

Good at

  • Fast inference on low‑memory GPUs
  • Small footprint for edge deployments
  • Supports 4096 token context

Limits

  • No vision or multimodal support
  • Requires at least 1 GB VRAM for Q2_K and higher quants
  • No mixture of experts

Best for

  • Chatbots on laptops
  • Rapid prototyping of LLM applications
  • Educational demos

Who it’s for. Developers and hobbyists who need a small, fast Llama model for local inference. The model is suitable for projects that require quick turnaround and low resource consumption.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ3_S0.2 GB 0.8 GB
Fits Get file →
IQ3_XS0.2 GB 0.8 GB
Fits Get file →
Q2_K0.2 GB 0.8 GB
Fits Get file →
Q3_K_S0.2 GB 0.8 GB
Fits Get file →
IQ3_M0.2 GB 0.8 GB
Fits Get file →
IQ4_XS0.2 GB 0.8 GB
Fits Get file →
Q3_K_M0.2 GB 0.8 GB
Fits Get file →
Q3_K_L0.2 GB 0.8 GB
Fits Get file →
Q4_K_S0.2 GB 0.9 GB
Fits Get file →
Q4_K_M0.2 GB 0.9 GB
Fits Get file →
Q5_K_S0.2 GB 0.9 GB
Fits Get file →
Q5_K_M0.2 GB 0.9 GB
Fits Get file →
Q6_K0.3 GB 1.0 GB
Fits Get file →
Q8_00.3 GB 1.0 GB
Fits Get file →
F160.6 GB 1.2 GB
Fits Get file →

For 8 GB VRAM, use Q4_K_M (0.22 GB file, 1 GB memory) or Q5_K_M (0.24 GB file, 1 GB memory). If you have 12 GB VRAM, Q6_K (0.30 GB file, 1.1 GB memory) or Q8_0 (0.32 GB file, 1.1 GB memory) are also viable. The F16 quant (0.6 GB file, 1.4 GB memory) requires more memory but offers higher precision.

What GPU do you need for LilMistral DPO 300M?

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: the best quant that fits is F16 (0.6 GB file, about 1.4 GB in use).

Which LilMistral DPO 300M quant should you choose?

Q4_K_M is the usual starting point for LilMistral DPO 300M: it balances file size and quality well. Here it is a 0.2 GB file.

The Q8_0 file is 0.3 GB, about 1.5 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is IQ3_S at 0.2 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters300M
Layers20
Trained context4,096 tokens
KV cache (fp16)35 KB per token

Source

Quantized bymradermacher
Licenseapache-2.0
Downloads (HF)645
Last updated2024-08-01

Frequently asked questions about LilMistral DPO 300M

How much VRAM does LilMistral DPO 300M need?

About 1.0 GB at Q4_K_M with an 8K context window. That is the 0.2 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 LilMistral DPO 300M on an 8 GB GPU?

Yes, with F16: the 0.6 GB file needs about 1.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of LilMistral DPO 300M GGUF?

It depends on the quant: IQ3_S is 0.2 GB, IQ3_XS is 0.2 GB, Q2_K is 0.2 GB, Q3_K_S is 0.2 GB, IQ3_M is 0.2 GB, IQ4_XS is 0.2 GB and 9 more.

What context length does LilMistral DPO 300M support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 0.3 GB at 8K and 0.1 GB at 4K.

Where can I download LilMistral DPO 300M GGUF?

The files are on Hugging Face in the mradermacher/LilMistral-DPO-300M-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 LilMistral DPO 300M use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run LilMistral DPO 300M on 8 GB VRAM?

Yes, the Q4_K_M quant uses about 1 GB VRAM, which fits comfortably in an 8 GB GPU. Other quants like Q5_K_M also fit within 8 GB.

Does the model support vision or multimodal inputs?

No, the model only processes text and has no vision or multimodal capabilities.

What license applies to this model?

The model is released under the apache-2.0 license.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

Other tiny models

All tiny models →