c4ai-command-r-plus-08-2024-IMat-GGUF

Llama.cpp imatrix quantization of CohereForAI/c4ai-command-r-plus-08-2024

Original Model: CohereForAI/c4ai-command-r-plus-08-2024
Original dtype: FP16 (float16)
Quantized by: llama.cpp b3645
IMatrix dataset: here

• Files
  • IMatrix
  • Common Quants
  • All Quants
• Downloading using huggingface-cli
• Inference
  • Simple chat template
  • Chat template with system prompt
  • Llama.cpp
• FAQ
  • Why is the IMatrix not applied everywhere?
  • How do I merge a split GGUF?

---

Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
c4ai-command-r-plus-08-2024.Q8_0	Q8_0	-	⏳ Processing	⚪ Static	-
c4ai-command-r-plus-08-2024.Q6_K/*	Q6_K	85.17GB	✅ Available	⚪ Static	✂ Yes
c4ai-command-r-plus-08-2024.Q4_K/*	Q4_K	62.75GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q3_K/*	Q3_K	50.98GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q2_K.gguf	Q2_K	39.50GB	✅ Available	🟢 IMatrix	📦 No

All Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
c4ai-command-r-plus-08-2024.FP16/*	F16	207.64GB	✅ Available	⚪ Static	✂ Yes
c4ai-command-r-plus-08-2024.Q8_0	Q8_0	-	⏳ Processing	⚪ Static	-
c4ai-command-r-plus-08-2024.Q6_K/*	Q6_K	85.17GB	✅ Available	⚪ Static	✂ Yes
c4ai-command-r-plus-08-2024.Q5_K/*	Q5_K	73.62GB	✅ Available	⚪ Static	✂ Yes
c4ai-command-r-plus-08-2024.Q5_K_S/*	Q5_K_S	71.80GB	✅ Available	⚪ Static	✂ Yes
c4ai-command-r-plus-08-2024.Q4_K/*	Q4_K	62.75GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q4_K_S/*	Q4_K_S	59.64GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.IQ4_NL/*	IQ4_NL	59.32GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.IQ4_XS/*	IQ4_XS	56.20GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q3_K/*	Q3_K	50.98GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q3_K_L/*	Q3_K_L	55.40GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.Q3_K_S/*	Q3_K_S	45.85GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.IQ3_M/*	IQ3_M	47.68GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.IQ3_S/*	IQ3_S	45.96GB	✅ Available	🟢 IMatrix	✂ Yes
c4ai-command-r-plus-08-2024.IQ3_XS.gguf	IQ3_XS	43.60GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ3_XXS.gguf	IQ3_XXS	40.66GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.Q2_K.gguf	Q2_K	39.50GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.Q2_K_S.gguf	Q2_K_S	36.60GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ2_M.gguf	IQ2_M	36.04GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ2_S.gguf	IQ2_S	33.32GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ2_XS.gguf	IQ2_XS	31.63GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ2_XXS.gguf	IQ2_XXS	28.61GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ1_M.gguf	IQ1_M	25.22GB	✅ Available	🟢 IMatrix	📦 No
c4ai-command-r-plus-08-2024.IQ1_S.gguf	IQ1_S	23.18GB	✅ Available	🟢 IMatrix	📦 No

Downloading using huggingface-cli

If you do not have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Download the specific file you want:

huggingface-cli download legraphista/c4ai-command-r-plus-08-2024-IMat-GGUF --include "c4ai-command-r-plus-08-2024.Q8_0.gguf" --local-dir ./

If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download legraphista/c4ai-command-r-plus-08-2024-IMat-GGUF --include "c4ai-command-r-plus-08-2024.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

---

Inference

Simple chat template

<BOS_TOKEN><|START_OF_TURN_TOKEN|><|USER_TOKEN|>{user_prompt}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>{assistant_response}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>{next_user_prompt}<|END_OF_TURN_TOKEN|>

Chat template with system prompt

<BOS_TOKEN><|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>{system_prompt}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>{user_prompt}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>{assistant_response}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>{next_user_prompt}<|END_OF_TURN_TOKEN|>

Llama.cpp

llama.cpp/main -m c4ai-command-r-plus-08-2024.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"

---

FAQ

Why is the IMatrix not applied everywhere?

According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).

How do I merge a split GGUF?

1. Make sure you have gguf-split available
   • To get hold of gguf-split, navigate to https://github.com/ggerganov/llama.cpp/releases
   • Download the appropriate zip for your system from the latest release
   • Unzip the archive and you should be able to find gguf-split
2. Locate your GGUF chunks folder (ex: c4ai-command-r-plus-08-2024.Q8_0)
3. Run gguf-split --merge c4ai-command-r-plus-08-2024.Q8_0/c4ai-command-r-plus-08-2024.Q8_0-00001-of-XXXXX.gguf c4ai-command-r-plus-08-2024.Q8_0.gguf
   • Make sure to point gguf-split to the first chunk of the split.

---

Got a suggestion? Ping me @legraphista!