You can deploy this model and start earning money today!

bartowski/
sophosympatheia_Glistening-Gem-31B-v2.1-GGUF

Name: sophosympatheia_Glistening-Gem-31B-v2.1-GGUF
Availability: InStock

gguf

mergekit

merge

not-for-all-audiences

any-to-any

sophosympatheia/Glistening-Gem-31B-v2.1

Views

No views yet

Llamacpp imatrix Quantizations of Glistening-Gem-31B-v2.1 by sophosympatheia

Using llama.cpp release b10419 for quantization.

Original model: https://huggingface.co/sophosympatheia/Glistening-Gem-31B-v2.1

Model details:

Parameter count: 31B
Input support: text, image and audio (with mmproj file) - details
Speculative decoding: no
imatrix: yes - details
Perplexity/KLD measured: no

How to run

Prompt format

<bos><|turn>system
{system_prompt}<turn|>
<|turn>user
{prompt}<turn|>
<|turn>model
<|channel>thought
<channel|>

Don't know which to choose? Grab Q4_K_M (19.60GB) - usually a good mix of size and performance. Download instructions available here

Available files:

Filename	Quant type	File Size	Split	Description
sophosympatheia_Glistening-Gem-31B-v2.1-bf16.gguf	bf16	61.41GB	true	Full BF16 weights.
sophosympatheia_Glistening-Gem-31B-v2.1-Q8_0.gguf	Q8_0	32.64GB	false	Extremely high quality, generally unneeded but max available quant.
sophosympatheia_Glistening-Gem-31B-v2.1-Q6_K_L.gguf	Q6_K_L	27.07GB	false	Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q6_K.gguf	Q6_K	26.73GB	false	Very high quality, near perfect, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q5_K_L.gguf	Q5_K_L	22.95GB	false	Uses Q8_0 for embed and output weights. High quality, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q5_K_M.gguf	Q5_K_M	22.61GB	false	High quality, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q5_K_S.gguf	Q5_K_S	21.50GB	false	High quality, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q4_K_L.gguf	Q4_K_L	19.94GB	false	Uses Q8_0 for embed and output weights. Good quality, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q4_1.gguf	Q4_1	19.77GB	false	Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon.
sophosympatheia_Glistening-Gem-31B-v2.1-Q4_K_M.gguf	Q4_K_M	19.60GB	false	Good quality, default size for most use cases, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q4_K_S.gguf	Q4_K_S	18.20GB	false	Slightly lower quality with more space savings, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q4_0.gguf	Q4_0	18.08GB	false	Legacy format, kept for compatibility with older tools.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ4_NL.gguf	IQ4_NL	18.03GB	false	Similar to IQ4_XS, but slightly larger.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ4_XS.gguf	IQ4_XS	17.16GB	false	Decent quality, smaller than Q4_K_S with similar performance, recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-Q3_K_XL.gguf	Q3_K_XL	17.15GB	false	Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability.
sophosympatheia_Glistening-Gem-31B-v2.1-Q3_K_L.gguf	Q3_K_L	16.81GB	false	Lower quality but usable, good for low RAM availability.
sophosympatheia_Glistening-Gem-31B-v2.1-Q3_K_M.gguf	Q3_K_M	15.92GB	false	Low quality.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ3_M.gguf	IQ3_M	15.13GB	false	Medium-low quality, new method with decent performance comparable to Q3_K_M.
sophosympatheia_Glistening-Gem-31B-v2.1-Q3_K_S.gguf	Q3_K_S	14.33GB	false	Low quality, not recommended.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ3_XS.gguf	IQ3_XS	13.84GB	false	Lower quality, new method with decent performance, slightly better than Q3_K_S.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ3_XXS.gguf	IQ3_XXS	12.98GB	false	Lower quality, new method with decent performance, comparable to Q3 quants.
sophosympatheia_Glistening-Gem-31B-v2.1-Q2_K_L.gguf	Q2_K_L	12.97GB	false	Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ2_M.gguf	IQ2_M	12.65GB	false	Relatively low quality, uses SOTA techniques to be surprisingly usable.
sophosympatheia_Glistening-Gem-31B-v2.1-Q2_K.gguf	Q2_K	12.63GB	false	Very low quality but surprisingly usable.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ2_S.gguf	IQ2_S	12.08GB	false	Low quality, uses SOTA techniques to be usable.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ2_XS.gguf	IQ2_XS	11.50GB	false	Low quality, uses SOTA techniques to be usable.
sophosympatheia_Glistening-Gem-31B-v2.1-IQ2_XXS.gguf	IQ2_XXS	10.83GB	false	Very low quality, uses SOTA techniques to be usable.

Download a specific file:

hf download bartowski/sophosympatheia_Glistening-Gem-31B-v2.1-GGUF --include "sophosympatheia_Glistening-Gem-31B-v2.1-Q4_K_M.gguf" --local-dir ./

Downloading using the Hugging Face CLI

Click to view download instructions

First, make sure you have the Hugging Face CLI installed:

pip install -U "huggingface_hub[cli]"

Download a specific file:

hf download bartowski/sophosympatheia_Glistening-Gem-31B-v2.1-GGUF --include "sophosympatheia_Glistening-Gem-31B-v2.1-Q4_K_M.gguf" --local-dir ./

The files marked true in the Split column above are stored as multiple parts in a folder. To download all the parts to a local folder, run:

hf download bartowski/sophosympatheia_Glistening-Gem-31B-v2.1-GGUF --include "sophosympatheia_Glistening-Gem-31B-v2.1-bf16/*" --local-dir ./

You can either specify a new local-dir (sophosympatheia_Glistening-Gem-31B-v2.1-bf16) or download them all in place (./)

How to run

These quants run with llama.cpp - installable in one line via llama.app:

curl -LsSf https://llama.app/install.sh | sh
llama-server -hf bartowski/sophosympatheia_Glistening-Gem-31B-v2.1-GGUF:Q4_K_M

llama-server includes a built-in chat web UI, served at http://localhost:8080 by default.

These quants were made with llama.cpp release b10419 - if this model's architecture is newly supported, you'll need that release or newer to run them.

They also work in: LM Studio · koboldcpp · ramalama · Jan AI · Text Generation Web UI · LoLLMs · Atomic Chat

Multimodal

This model supports image and audio input. Alongside the quants, this repo includes the multimodal projector files mmproj-sophosympatheia_Glistening-Gem-31B-v2.1-f16.gguf and mmproj-sophosympatheia_Glistening-Gem-31B-v2.1-bf16.gguf, which pair with any quant above.

llama.cpp downloads the mmproj automatically when using -hf as shown above; if you're loading files manually, pass it with --mmproj.

imatrix

All quants made using imatrix option, with a calibration corpus rendered through this model's own chat template. The corpus pairs plain prose with tool-calling and reasoning conversations (corpus source data), encoded exactly as this model sees them at inference and processed with --parse-special, so chat-format special tokens contribute to the importance matrix. The corpus rendered for this model is included in this repo: sophosympatheia_Glistening-Gem-31B-v2.1-calibration-v6.txt. The imatrix is available here: sophosympatheia_Glistening-Gem-31B-v2.1-imatrix.gguf.

Calibration render details

json

{
  "generator": "auto_quant_v2 calibration renderer",
  "recipe": "calibration-v6",
  "model": "Glistening-Gem-31B-v2.1",
  "encoder": "chat_template",
  "chunk_size": 512,
  "prose_chunks": 219,
  "tool_chunks": 273,
  "total_chunks": 492,
  "tool_chunk_fraction": 0.555,
  "n_conversations": 137,
  "extension_convs_used": 0,
  "conversation_token_lengths": [
    345,
    1411,
    652,
    1146,
    879,
    1186,
    2285,
    410,
    987,
    945,
    722,
    1728,
    667,
    739,
    2132,
    993,
    836,
    713,
    552,
    518,
    1152,
    782,
    1143,
    1012,
    1435,
    1215,
    797,
    470,
    752,
    1242,
    1309,
    590,
    1027,
    761,
    986,
    1458,
    1413,
    928,
    284,
    1746,
    1225,
    888,
    1157,
    1491,
    1715,
    1046,
    1269,
    518,
    2154,
    909,
    2418,
    569,
    761,
    515,
    919,
    491,
    2037,
    498,
    871,
    821,
    1008,
    963,
    680,
    772,
    742,
    1163,
    669,
    1272,
    1562,
    660,
    329,
    881,
    2853,
    2736,
    529,
    505,
    775,
    550,
    1061,
    869,
    921,
    526,
    807,
    826,
    926,
    1234,
    1158,
    1172,
    612,
    425,
    2180,
    660,
    907,
    1319,
    1790,
    972,
    417,
    1191,
    1133,
    1381,
    1394,
    1415,
    591,
    534,
    814,
    2647,
    530,
    357,
    561,
    982,
    872,
    1107,
    539,
    361,
    332,
    2029,
    496,
    923,
    1593,
    1438,
    2082,
    2110,
    459,
    670,
    615,
    600,
    1009,
    769,
    646,
    1093,
    547,
    517,
    1170,
    710,
    688,
    1087,
    1257
  ],
  "warnings": []
}

Embed/output weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

ARM/AVX information

llama.cpp automatically "repacks" weights into an interleaved layout at load time for faster inference on ARM and AVX machines - details in this PR. This once required downloading special Q4_0_4_4/4_8/8_8 files; those are long gone. Online repacking now covers Q4_0, IQ4_NL, and most K-quants, so no special quant choice is needed for CPU inference.

Which file should I choose?

Click here for details

An older (early 2024) but still useful write-up with charts comparing quant performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.

Hugging Face can also do this math for you: add your hardware in your Local Apps settings and the model page will show which files fit.

Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.

If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.

If you want to get more into the weeds, you can check out this extremely useful feature chart:

llama.cpp feature matrix

But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.

These I-quants can also be used on CPU, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.

Credits

Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.

Thank you ZeroWw for the inspiration to experiment with embed/output.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski