legraphista's picture
Upload README.md with huggingface_hub
c294fd9 verified
|
raw
history blame
9.85 kB
metadata
base_model: Qwen/Qwen2.5-0.5B-Instruct
inference: false
language:
  - en
library_name: gguf
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct/blob/main/LICENSE
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - chat
  - quantized
  - GGUF
  - quantization
  - imat
  - imatrix
  - static
  - 16bit
  - 8bit
  - 6bit
  - 5bit
  - 4bit
  - 3bit
  - 2bit
  - 1bit

Qwen2.5-0.5B-Instruct-IMat-GGUF

Llama.cpp imatrix quantization of Qwen/Qwen2.5-0.5B-Instruct

Original Model: Qwen/Qwen2.5-0.5B-Instruct
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3785
IMatrix dataset: here


Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Qwen2.5-0.5B-Instruct.Q8_0.gguf Q8_0 531.07MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q6_K.gguf Q6_K 505.74MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q4_K.gguf Q4_K 397.81MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q3_K.gguf Q3_K 355.47MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q2_K.gguf Q2_K 338.61MB βœ… Available 🟒 IMatrix πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Qwen2.5-0.5B-Instruct.BF16.gguf BF16 994.16MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.FP16.gguf F16 994.16MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q8_0.gguf Q8_0 531.07MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q6_K.gguf Q6_K 505.74MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q5_K.gguf Q5_K 420.09MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q5_K_S.gguf Q5_K_S 412.71MB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q4_K.gguf Q4_K 397.81MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q4_K_S.gguf Q4_K_S 385.47MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ4_NL.gguf IQ4_NL 352.67MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ4_XS.gguf IQ4_XS 349.40MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q3_K.gguf Q3_K 355.47MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q3_K_L.gguf Q3_K_L 369.36MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q3_K_S.gguf Q3_K_S 338.26MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ3_M.gguf IQ3_M 342.75MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ3_S.gguf IQ3_S 338.61MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ3_XS.gguf IQ3_XS 338.61MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ3_XXS.gguf IQ3_XXS 333.70MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q2_K.gguf Q2_K 338.61MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.Q2_K_S.gguf Q2_K_S 331.05MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ2_M.gguf IQ2_M 328.60MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ2_S.gguf IQ2_S 325.74MB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-0.5B-Instruct.IQ2_XS IQ2_XS - ⏳ Processing 🟒 IMatrix -
Qwen2.5-0.5B-Instruct.IQ2_XXS IQ2_XXS - ⏳ Processing 🟒 IMatrix -
Qwen2.5-0.5B-Instruct.IQ1_M IQ1_M - ⏳ Processing 🟒 IMatrix -
Qwen2.5-0.5B-Instruct.IQ1_S IQ1_S - ⏳ Processing 🟒 IMatrix -

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/Qwen2.5-0.5B-Instruct-IMat-GGUF --include "Qwen2.5-0.5B-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/Qwen2.5-0.5B-Instruct-IMat-GGUF --include "Qwen2.5-0.5B-Instruct.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<|im_start|>system
You are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>
<|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 Qwen2.5-0.5B-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
  2. Locate your GGUF chunks folder (ex: Qwen2.5-0.5B-Instruct.Q8_0)
  3. Run gguf-split --merge Qwen2.5-0.5B-Instruct.Q8_0/Qwen2.5-0.5B-Instruct.Q8_0-00001-of-XXXXX.gguf Qwen2.5-0.5B-Instruct.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!