dolphin-2.9.2-Phi-3-Medium-IMat-GGUF

Llama.cpp imatrix quantization of cognitivecomputations/dolphin-2.9.2-Phi-3-Medium

Original Model: cognitivecomputations/dolphin-2.9.2-Phi-3-Medium
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3072
IMatrix dataset: here


Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
dolphin-2.9.2-Phi-3-Medium.Q8_0.gguf Q8_0 14.83GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q6_K.gguf Q6_K 11.45GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q4_K.gguf Q4_K 8.40GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q3_K.gguf Q3_K 6.75GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q2_K.gguf Q2_K 5.20GB βœ… Available 🟒 IMatrix πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
dolphin-2.9.2-Phi-3-Medium.BF16.gguf BF16 27.92GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.FP16.gguf F16 27.92GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q8_0.gguf Q8_0 14.83GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q6_K.gguf Q6_K 11.45GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q5_K.gguf Q5_K 9.88GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q5_K_S.gguf Q5_K_S 9.62GB βœ… Available βšͺ Static πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q4_K.gguf Q4_K 8.40GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q4_K_S.gguf Q4_K_S 7.96GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ4_NL.gguf IQ4_NL 7.93GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ4_XS.gguf IQ4_XS 7.51GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q3_K.gguf Q3_K 6.75GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q3_K_L.gguf Q3_K_L 7.35GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q3_K_S.gguf Q3_K_S 6.06GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ3_M.gguf IQ3_M 6.30GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ3_S.gguf IQ3_S 6.10GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ3_XS.gguf IQ3_XS 5.78GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ3_XXS.gguf IQ3_XXS 5.42GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q2_K.gguf Q2_K 5.20GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.Q2_K_S.gguf Q2_K_S 4.83GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ2_M.gguf IQ2_M 4.78GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ2_S.gguf IQ2_S 4.41GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ2_XS.gguf IQ2_XS 4.20GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ2_XXS.gguf IQ2_XXS 3.80GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ1_M.gguf IQ1_M 3.33GB βœ… Available 🟒 IMatrix πŸ“¦ No
dolphin-2.9.2-Phi-3-Medium.IQ1_S.gguf IQ1_S 3.05GB βœ… Available 🟒 IMatrix πŸ“¦ 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/dolphin-2.9.2-Phi-3-Medium-IMat-GGUF --include "dolphin-2.9.2-Phi-3-Medium.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/dolphin-2.9.2-Phi-3-Medium-IMat-GGUF --include "dolphin-2.9.2-Phi-3-Medium.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Chat template with system prompt

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Llama.cpp

llama.cpp/main -m dolphin-2.9.2-Phi-3-Medium.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
  2. Locate your GGUF chunks folder (ex: dolphin-2.9.2-Phi-3-Medium.Q8_0)
  3. Run gguf-split --merge dolphin-2.9.2-Phi-3-Medium.Q8_0/dolphin-2.9.2-Phi-3-Medium.Q8_0-00001-of-XXXXX.gguf dolphin-2.9.2-Phi-3-Medium.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!

Downloads last month
789
GGUF
Model size
14B params
Architecture
llama

1-bit

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

16-bit

Inference Examples
Inference API (serverless) has been turned off for this model.

Model tree for legraphista/dolphin-2.9.2-Phi-3-Medium-IMat-GGUF

Datasets used to train legraphista/dolphin-2.9.2-Phi-3-Medium-IMat-GGUF