Phi-3-mini-4k-instruct-IMat-GGUF

Llama.cpp imatrix quantization of microsoft/Phi-3-mini-4k-instruct

Original Model: microsoft/Phi-3-mini-4k-instruct
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3001
IMatrix dataset: here

• Phi-3-mini-4k-instruct-IMat-GGUF
  • Files
    • IMatrix
    • Common Quants
    • All Quants
  • Downloading using huggingface-cli
  • Inference
    • Simple chat template
    • 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
Phi-3-mini-4k-instruct.Q8_0.gguf	Q8_0	4.06GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.Q6_K.gguf	Q6_K	3.14GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.Q4_K.gguf	Q4_K	2.39GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.Q3_K.gguf	Q3_K	1.96GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.Q2_K.gguf	Q2_K	1.42GB	✅ Available	🟢 Yes	📦 No

All Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
Phi-3-mini-4k-instruct.FP16.gguf	F16	7.64GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.BF16.gguf	BF16	7.64GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.Q5_K.gguf	Q5_K	2.82GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.Q5_K_S.gguf	Q5_K_S	2.64GB	✅ Available	⚪ No	📦 No
Phi-3-mini-4k-instruct.Q4_K_S.gguf	Q4_K_S	2.19GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.Q3_K_L.gguf	Q3_K_L	2.09GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.Q3_K_S.gguf	Q3_K_S	1.68GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.Q2_K_S.gguf	Q2_K_S	1.33GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ4_NL.gguf	IQ4_NL	2.18GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ4_XS.gguf	IQ4_XS	2.06GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ3_M.gguf	IQ3_M	1.86GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ3_S.gguf	IQ3_S	1.68GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ3_XS.gguf	IQ3_XS	1.63GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ3_XXS.gguf	IQ3_XXS	1.51GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ2_M.gguf	IQ2_M	1.32GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ2_S.gguf	IQ2_S	1.22GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ2_XS.gguf	IQ2_XS	1.15GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ2_XXS.gguf	IQ2_XXS	1.04GB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ1_M.gguf	IQ1_M	917.11MB	✅ Available	🟢 Yes	📦 No
Phi-3-mini-4k-instruct.IQ1_S.gguf	IQ1_S	841.61MB	✅ Available	🟢 Yes	📦 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/Phi-3-mini-4k-instruct-IMat-GGUF --include "Phi-3-mini-4k-instruct.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/Phi-3-mini-4k-instruct-IMat-GGUF --include "Phi-3-mini-4k-instruct.Q8_0/*" --local-dir Phi-3-mini-4k-instruct.Q8_0
# see FAQ for merging GGUF's

---

Inference

Simple chat template

<s><|user|>
I am going to Paris, what should I see?<|end|>
<|assistant|>
Paris, the capital of France, is known for its stunning architecture, art museums, historical landmarks, and romantic atmosphere. Here are some of the top attractions to see in Paris:

1. The Eiffel Tower: The iconic Eiffel Tower is one of the most recognizable landmarks in the world and offers breathtaking views of the city.
2. The Louvre Museum: The Louvre is one of the world's largest and most famous museums, housing an impressive collection of art and artifacts, including the Mona Lisa.
3. Notre-Dame Cathedral: This beautiful cathedral is one of the most famous landmarks in Paris and is known for its Gothic architecture and stunning stained glass windows.

These are just a few of the many attractions that Paris has to offer. With so much to see and do, it's no wonder that Paris is one of the most popular tourist destinations in the world."<|end|>
<|user|>
What is so great about #1?<|end|>
<|assistant|>

Llama.cpp

llama.cpp/main -m Phi-3-mini-4k-instruct.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: Phi-3-mini-4k-instruct.Q8_0)
3. Run gguf-split --merge Phi-3-mini-4k-instruct.Q8_0/Phi-3-mini-4k-instruct.Q8_0-00001-of-XXXXX.gguf Phi-3-mini-4k-instruct.Q8_0.gguf
   • Make sure to point gguf-split to the first chunk of the split.

---

Got a suggestion? Ping me @legraphista!