PU

Puma 3B i1

PumaGGUF by mradermacherUpdated 2024-11-02
Download on Hugging Face
Parameters3.43B
Context2K
Layers26
Architecturellama
Smallest quant2.0 GB
Quick answer

Puma 3B i1 needs about 5.8 GB of memory at Q4_K_M (a 2.6 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 21 quants, from 2.0 GB (IQ1_M) to 3.6 GB (Q6_K).

Puma 3B i1 – 3.43B llama model in GGUF format

Puma 3B i1 is a 3.43B parameter llama model finetuned on ShareGPT Hyperfiltered. It supports a 2048 token context window and is released under the apache-2.0 license.

Good at

  • Fine‑tuned conversational tasks
  • Short‑form content generation
  • Prompt‑based instruction following

Limits

  • 2048 token context
  • Requires 5.2–6.9 GB VRAM for 8k context
  • No vision or multimodal support

Best for

  • Chatbot prototypes
  • Quick research summaries
  • Lightweight local inference

Who it’s for. Developers and researchers needing a small llama model for local deployment. The model is suitable for experimentation and low‑resource inference environments.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_M2.0 GB 3.2 GB
Fits Get file →
IQ1_S2.0 GB 3.2 GB
Fits Get file →
IQ2_M2.0 GB 3.2 GB
Fits Get file →
IQ2_S2.0 GB 3.2 GB
Fits Get file →
IQ2_XS2.0 GB 3.2 GB
Fits Get file →
IQ2_XXS2.0 GB 3.2 GB
Fits Get file →
IQ3_S2.0 GB 3.2 GB
Fits Get file →
IQ3_XS2.0 GB 3.2 GB
Fits Get file →
IQ3_XXS2.0 GB 3.2 GB
Fits Get file →
IQ4_XS2.0 GB 3.2 GB
Fits Get file →
Q2_K2.0 GB 3.2 GB
Fits Get file →
Q3_K_S2.0 GB 3.2 GB
Fits Get file →
Q4_02.0 GB 3.2 GB
Fits Get file →
IQ3_M2.1 GB 3.2 GB
Fits Get file →
Q3_K_M2.1 GB 3.3 GB
Fits Get file →
Q3_K_L2.2 GB 3.4 GB
Fits Get file →
Q4_K_S2.4 GB 3.6 GB
Fits Get file →
Q4_K_M2.6 GB 3.8 GB
Fits Get file →
Q5_K_S2.6 GB 3.8 GB
Fits Get file →
Q5_K_M2.8 GB 3.9 GB
Fits Get file →
Q6_K3.6 GB 4.8 GB
Fits Get file →

For 8 GB VRAM, use Q2_K or IQ1_M (1.98 GB file, ~5.2 GB VRAM). For 12 GB VRAM, Q4_K_M (2.58 GB file, ~5.8 GB VRAM) is a good balance. Higher quants like Q6_K (3.64 GB file, ~6.9 GB VRAM) offer more precision if memory allows.

What GPU do you need for Puma 3B i1?

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: the best quant that fits is Q5_K_M (2.8 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K (3.6 GB file, about 6.9 GB in use).

Which Puma 3B i1 quant should you choose?

Q4_K_M is the usual starting point for Puma 3B i1: it balances file size and quality well. Here it is a 2.6 GB file.

If memory is tight, the smallest file is IQ1_M at 2.0 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

Architecturellama
Parameters3.43B
Layers26
Trained context2,048 tokens
KV cache (fp16)325 KB per token

Source

Quantized bymradermacher
FamilyPuma
Base modelacrastt/Puma-3B
Licenseapache-2.0
Downloads (HF)566
Last updated2024-11-02

Frequently asked questions about Puma 3B i1

How much VRAM does Puma 3B i1 need?

About 5.8 GB at Q4_K_M with an 8K context window. That is the 2.6 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 Puma 3B i1 on an 8 GB GPU?

Yes, with Q6_K: the 3.6 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Puma 3B i1 GGUF?

It depends on the quant: IQ1_M is 2.0 GB, IQ1_S is 2.0 GB, IQ2_M is 2.0 GB, IQ2_S is 2.0 GB, IQ2_XS is 2.0 GB, IQ2_XXS is 2.0 GB and 15 more.

What context length does Puma 3B i1 support?

The GGUF header lists a trained context length of 2,048 tokens (2K). The KV cache grows with the context you use: about 2.7 GB at 8K and 0.7 GB at 2K.

Where can I download Puma 3B i1 GGUF?

The files are on Hugging Face in the mradermacher/Puma-3B-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 Puma 3B i1 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 Puma 3B i1 on 8 GB VRAM?

Yes, the Q2_K and IQ1_M quantizations are 1.98 GB and need about 5.2 GB VRAM at 8k context, fitting within 8 GB.

Does this model support vision?

No, Puma 3B i1 is a text‑only llama model with no vision capabilities.

What license applies to this model?

The model is released under the apache-2.0 license.

Other GGUFs of Puma-3B

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Puma models

All Puma →

Other tiny models

All tiny models →