Hugging Face's logo

legraphista
/
openchat-3.6-8b-20240522-IMat-GGUF

Text Generation
GGUF
quantized
GGUF
imatrix
quantization
imat
static
conversational
Model card Files Community
openchat-3.6-8b-20240522-IMat-GGUF / README.md
legraphista's picture
legraphista
Upload README.md with huggingface_hub
f3bbb21 verified
preview code
|
raw
history blame contribute delete
10.5 kB
metadata
base_model: openchat/openchat-3.6-8b-20240522
inference: false
library_name: gguf
license: llama3
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - quantized
  - GGUF
  - imatrix
  - quantization
  - imat
  - imatrix
  - static

openchat-3.6-8b-20240522-IMat-GGUF

Llama.cpp imatrix quantization of openchat/openchat-3.6-8b-20240522

Original Model: openchat/openchat-3.6-8b-20240522
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3006
IMatrix dataset: here

Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
openchat-3.6-8b-20240522.Q8_0.gguf Q8_0 8.54GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.Q6_K.gguf Q6_K 6.60GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.Q4_K.gguf Q4_K 4.92GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.Q3_K.gguf Q3_K 4.02GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.Q2_K.gguf Q2_K 3.18GB ✅ Available 🟢 IMatrix 📦 No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
openchat-3.6-8b-20240522.FP16.gguf F16 16.07GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.BF16.gguf BF16 16.07GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.Q5_K.gguf Q5_K 5.73GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.Q5_K_S.gguf Q5_K_S 5.60GB ✅ Available ⚪ Static 📦 No
openchat-3.6-8b-20240522.Q4_K_S.gguf Q4_K_S 4.69GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.Q3_K_L.gguf Q3_K_L 4.32GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.Q3_K_S.gguf Q3_K_S 3.66GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.Q2_K_S.gguf Q2_K_S 2.99GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ4_NL.gguf IQ4_NL 4.68GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ4_XS.gguf IQ4_XS 4.45GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ3_M.gguf IQ3_M 3.78GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ3_S.gguf IQ3_S 3.68GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ3_XS.gguf IQ3_XS 3.52GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ3_XXS.gguf IQ3_XXS 3.27GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ2_M.gguf IQ2_M 2.95GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ2_S.gguf IQ2_S 2.76GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ2_XS.gguf IQ2_XS 2.61GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ2_XXS.gguf IQ2_XXS 2.40GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ1_M.gguf IQ1_M 2.16GB ✅ Available 🟢 IMatrix 📦 No
openchat-3.6-8b-20240522.IQ1_S.gguf IQ1_S 2.02GB ✅ 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/openchat-3.6-8b-20240522-IMat-GGUF --include "openchat-3.6-8b-20240522.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/openchat-3.6-8b-20240522-IMat-GGUF --include "openchat-3.6-8b-20240522.Q8_0/*" --local-dir openchat-3.6-8b-20240522.Q8_0
# see FAQ for merging GGUF's

Inference

Simple chat template 

<|begin_of_text|><|start_header_id|>GPT4 Correct User<|end_header_id|>

Can you provide ways to eat combinations of bananas and dragonfruits?<|eot_id|><|start_header_id|>GPT4 Correct Assistant<|end_header_id|>

Sure! Here are some ways to eat bananas and dragonfruits together:
 1. Banana and dragonfruit smoothie: Blend bananas and dragonfruits together with some milk and honey.
 2. Banana and dragonfruit salad: Mix sliced bananas and dragonfruits together with some lemon juice and honey.<|eot_id|><|start_header_id|>GPT4 Correct User<|end_header_id|>

What about solving an 2x + 3 = 7 equation?<|eot_id|>

Chat template with system prompt 

<|begin_of_text|><|start_header_id|>System<|end_header_id|>

You are a helpful AI.<|eot_id|><|start_header_id|>GPT4 Correct User<|end_header_id|>

Can you provide ways to eat combinations of bananas and dragonfruits?<|eot_id|><|start_header_id|>GPT4 Correct Assistant<|end_header_id|>

Sure! Here are some ways to eat bananas and dragonfruits together:
 1. Banana and dragonfruit smoothie: Blend bananas and dragonfruits together with some milk and honey.
 2. Banana and dragonfruit salad: Mix sliced bananas and dragonfruits together with some lemon juice and honey.<|eot_id|><|start_header_id|>GPT4 Correct User<|end_header_id|>

What about solving an 2x + 3 = 7 equation?<|eot_id|>

Llama.cpp 

llama.cpp/main -m openchat-3.6-8b-20240522.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: openchat-3.6-8b-20240522.Q8_0)
  3. Run gguf-split --merge openchat-3.6-8b-20240522.Q8_0/openchat-3.6-8b-20240522.Q8_0-00001-of-XXXXX.gguf openchat-3.6-8b-20240522.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!