CO

c4ai-command-r-plus

Command RGGUF by qwp4w3hybUpdated 2024-05-29
Download on Hugging Face
Parameters104B
Context128K
Layers64
Architecturecommand-r
Smallest quant23.2 GB
Quick answer

c4ai-command-r-plus needs about 65.4 GB of memory at Q4_K_M (a 62.8 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 19 quants, from 23.2 GB (IQ1_S) to 207.6 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.

C4AI Command R+ 104B parameter model, quantized for llama.cpp

C4AI Command R+ is a 104B parameter model with Retrieval Augmented Generation and tool use. It is available in multiple quantized forms for llama.cpp. The model is licensed under cc‑by‑nc‑4.0.

Good at

  • Retrieval Augmented Generation
  • Multi‑step tool use
  • Large context handling

Limits

  • No vision support
  • Requires large VRAM for high‑quality quants
  • License restricts commercial use

Best for

  • Research on RAG systems
  • Tool‑enabled task automation
  • Large‑context text generation

Who it’s for. Researchers and developers needing advanced RAG and tool use capabilities in a large language model. The model is suitable for non‑commercial projects due to its license.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S23.2 GB 25.8 GB
Too big Get file →
IQ2_XXS28.6 GB 31.3 GB
Too big Get file →
IQ2_XS31.6 GB 34.3 GB
Too big Get file →
IQ2_S33.3 GB 36.0 GB
Too big Get file →
IQ2_M36.0 GB 38.7 GB
Too big Get file →
IQ3_XXS40.7 GB 43.3 GB
Too big Get file →
IQ3_XS43.6 GB 46.2 GB
Too big Get file →
IQ3_S46.0 GB 48.6 GB
Too big Get file →
IQ3_M47.7 GB 50.3 GB
Too big Get file →
IQ4_XS56.2 GB (2 parts) 58.8 GB
Too big Get file →
IQ4_NL59.3 GB (2 parts) 62.0 GB
Too big Get file →
Q4_059.4 GB (2 parts) 62.1 GB
Too big Get file →
Q4_K_S59.6 GB (2 parts) 62.3 GB
Too big Get file →
Q4_K_M62.8 GB (2 parts) 65.4 GB
Too big Get file →
Q5_K_S71.8 GB (2 parts) 74.5 GB
Too big Get file →
Q5_K_M73.6 GB (2 parts) 76.3 GB
Too big Get file →
Q6_K85.2 GB (2 parts) 87.8 GB
Too big Get file →
Q8_0110.3 GB (3 parts) 113.0 GB
Too big Get file →
F16207.6 GB (5 parts) 210.3 GB
Too big Get file →

For 8‑GB VRAM, use IQ1_S (23.18 GB file, 25.8 GB memory at 8k context). For 16‑GB VRAM, IQ2_XS (31.63 GB file, 34.3 GB memory). For 24‑GB VRAM, Q4_K_M split into 2 shards (62.75 GB file, 65.4 GB memory).

What GPU do you need for c4ai-command-r-plus?

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 (28.6 GB file, about 31.3 GB in use).
  • 48 GB VRAM: the best quant that fits is IQ3_XS (43.6 GB file, about 46.2 GB in use).

Which c4ai-command-r-plus quant should you choose?

Q4_K_M is the usual starting point for c4ai-command-r-plus: it balances file size and quality well. Here it is a 62.8 GB file.

The Q8_0 file is 110.3 GB, about 1.8 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 IQ1_S at 23.2 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

Architecturecommand-r
Parameters104B
Layers64
Trained context131,072 tokens
KV cache (fp16)256 KB per token

Source

Quantized byqwp4w3hyb
FamilyCommand R
Downloads (HF)558
Last updated2024-05-29

Frequently asked questions about c4ai-command-r-plus

How much VRAM does c4ai-command-r-plus need?

About 65.4 GB at Q4_K_M with an 8K context window. That is the 62.8 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 c4ai-command-r-plus on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 25.8 GB, so you would need a larger GPU or CPU offloading.

What is the file size of c4ai-command-r-plus GGUF?

It depends on the quant: IQ1_S is 23.2 GB, IQ2_XXS is 28.6 GB, IQ2_XS is 31.6 GB, IQ2_S is 33.3 GB, IQ2_M is 36.0 GB, IQ3_XXS is 40.7 GB and 13 more.

What context length does c4ai-command-r-plus support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.6 GB at 32K.

Where can I download c4ai-command-r-plus GGUF?

The files are on Hugging Face in the qwp4w3hyb/c4ai-command-r-plus-iMat-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 c4ai-command-r-plus 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.

Can I run C4AI Command R+ on 8 GB VRAM?

Use the IQ1_S quant, which requires 25.8 GB memory at 8k context, so 8 GB VRAM is insufficient.

Is the model suitable for commercial deployment?

The cc‑by‑nc‑4.0 license prohibits commercial use.

Does the model support vision tasks?

No, the model does not include vision capabilities.

Other GGUFs of c4ai-command-r-plus

Similar models

Browse all →

More from qwp4w3hyb

All by qwp4w3hyb →

More Command R models

All Command R →

Other large models

All large models →