From 648fc1f8ac84731961ae14faf95ef0084dd041d6 Mon Sep 17 00:00:00 2001 From: ModelHub XC Date: Sun, 21 Jun 2026 03:26:12 +0800 Subject: [PATCH] =?UTF-8?q?=E5=88=9D=E5=A7=8B=E5=8C=96=E9=A1=B9=E7=9B=AE?= =?UTF-8?q?=EF=BC=8C=E7=94=B1ModelHub=20XC=E7=A4=BE=E5=8C=BA=E6=8F=90?= =?UTF-8?q?=E4=BE=9B=E6=A8=A1=E5=9E=8B?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Model: bartowski/ibm-granite_granite-4.1-8b-GGUF Source: Original Platform --- .gitattributes | 85 +++++++++++ README.md | 180 ++++++++++++++++++++++++ configuration.json | 1 + ibm-granite_granite-4.1-8b-IQ2_M.gguf | 3 + ibm-granite_granite-4.1-8b-IQ3_M.gguf | 3 + ibm-granite_granite-4.1-8b-IQ3_XS.gguf | 3 + ibm-granite_granite-4.1-8b-IQ3_XXS.gguf | 3 + ibm-granite_granite-4.1-8b-IQ4_NL.gguf | 3 + ibm-granite_granite-4.1-8b-IQ4_XS.gguf | 3 + ibm-granite_granite-4.1-8b-Q2_K.gguf | 3 + ibm-granite_granite-4.1-8b-Q2_K_L.gguf | 3 + ibm-granite_granite-4.1-8b-Q3_K_L.gguf | 3 + ibm-granite_granite-4.1-8b-Q3_K_M.gguf | 3 + ibm-granite_granite-4.1-8b-Q3_K_S.gguf | 3 + ibm-granite_granite-4.1-8b-Q3_K_XL.gguf | 3 + ibm-granite_granite-4.1-8b-Q4_0.gguf | 3 + ibm-granite_granite-4.1-8b-Q4_1.gguf | 3 + ibm-granite_granite-4.1-8b-Q4_K_L.gguf | 3 + ibm-granite_granite-4.1-8b-Q4_K_M.gguf | 3 + ibm-granite_granite-4.1-8b-Q4_K_S.gguf | 3 + ibm-granite_granite-4.1-8b-Q5_K_L.gguf | 3 + ibm-granite_granite-4.1-8b-Q5_K_M.gguf | 3 + ibm-granite_granite-4.1-8b-Q5_K_S.gguf | 3 + ibm-granite_granite-4.1-8b-Q6_K.gguf | 3 + ibm-granite_granite-4.1-8b-Q6_K_L.gguf | 3 + ibm-granite_granite-4.1-8b-Q8_0.gguf | 3 + ibm-granite_granite-4.1-8b-bf16.gguf | 3 + ibm-granite_granite-4.1-8b-imatrix.gguf | 3 + 28 files changed, 341 insertions(+) create mode 100644 .gitattributes create mode 100644 README.md create mode 100644 configuration.json create mode 100644 ibm-granite_granite-4.1-8b-IQ2_M.gguf create mode 100644 ibm-granite_granite-4.1-8b-IQ3_M.gguf create mode 100644 ibm-granite_granite-4.1-8b-IQ3_XS.gguf create mode 100644 ibm-granite_granite-4.1-8b-IQ3_XXS.gguf create mode 100644 ibm-granite_granite-4.1-8b-IQ4_NL.gguf create mode 100644 ibm-granite_granite-4.1-8b-IQ4_XS.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q2_K.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q2_K_L.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q3_K_L.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q3_K_M.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q3_K_S.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q3_K_XL.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q4_0.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q4_1.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q4_K_L.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q4_K_M.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q4_K_S.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q5_K_L.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q5_K_M.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q5_K_S.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q6_K.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q6_K_L.gguf create mode 100644 ibm-granite_granite-4.1-8b-Q8_0.gguf create mode 100644 ibm-granite_granite-4.1-8b-bf16.gguf create mode 100644 ibm-granite_granite-4.1-8b-imatrix.gguf diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..1469218 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,85 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bin.* filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zstandard filter=lfs diff=lfs merge=lfs -text +*.tfevents* filter=lfs diff=lfs merge=lfs -text +*.db* filter=lfs diff=lfs merge=lfs -text +*.ark* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text +**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text + +*.ggml filter=lfs diff=lfs merge=lfs -text +*.llamafile* filter=lfs diff=lfs merge=lfs -text +*.pt2 filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-IQ3_XS.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-IQ4_NL.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q4_1.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q5_K_L.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-Q6_K_L.gguf filter=lfs diff=lfs merge=lfs -text + +ibm-granite_granite-4.1-8b-bf16.gguf filter=lfs diff=lfs merge=lfs -text +ibm-granite_granite-4.1-8b-imatrix.gguf filter=lfs diff=lfs merge=lfs -text \ No newline at end of file diff --git a/README.md b/README.md new file mode 100644 index 0000000..032b731 --- /dev/null +++ b/README.md @@ -0,0 +1,180 @@ +--- +quantized_by: bartowski +pipeline_tag: text-generation +base_model: ibm-granite/granite-4.1-8b +license: apache-2.0 +tags: +- language +- granite-4.1 +base_model_relation: quantized +--- + +## Llamacpp imatrix Quantizations of granite-4.1-8b by ibm-granite + +Using llama.cpp release b8970 for quantization. + +Original model: https://huggingface.co/ibm-granite/granite-4.1-8b + +All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/82ae9b520227f57d79ba04add13d0d0d) + +Run them in your choice of tools: + +- [llama.cpp](https://github.com/ggml-org/llama.cpp) +- [ramalama](https://github.com/containers/ramalama) +- [LM Studio](https://lmstudio.ai/) +- [koboldcpp](https://github.com/LostRuins/koboldcpp) +- [Jan AI](https://www.jan.ai/) +- [Text Generation Web UI](https://github.com/oobabooga/text-generation-webui) +- [LoLLMs](https://github.com/ParisNeo/lollms) + +Note: if it's a newly supported model, you may need to wait for an update from the developers. + +## Prompt format + +``` +<|start_of_role|>system<|end_of_role|>{system_prompt}<|end_of_text|> +<|start_of_role|>user<|end_of_role|>{prompt}<|end_of_text|> +<|start_of_role|>assistant<|end_of_role|> +``` + +## Download a file (not the whole branch) from below: + +| Filename | Quant type | File Size | Split | Description | +| -------- | ---------- | --------- | ----- | ----------- | +| [granite-4.1-8b-bf16.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-bf16.gguf) | bf16 | 17.59GB | false | Full BF16 weights. | +| [granite-4.1-8b-Q8_0.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q8_0.gguf) | Q8_0 | 9.35GB | false | Extremely high quality, generally unneeded but max available quant. | +| [granite-4.1-8b-Q6_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q6_K_L.gguf) | Q6_K_L | 7.72GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. | +| [granite-4.1-8b-Q6_K.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q6_K.gguf) | Q6_K | 7.52GB | false | Very high quality, near perfect, *recommended*. | +| [granite-4.1-8b-Q5_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q5_K_L.gguf) | Q5_K_L | 6.66GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. | +| [granite-4.1-8b-Q5_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q5_K_M.gguf) | Q5_K_M | 6.41GB | false | High quality, *recommended*. | +| [granite-4.1-8b-Q5_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q5_K_S.gguf) | Q5_K_S | 6.15GB | false | High quality, *recommended*. | +| [granite-4.1-8b-Q4_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q4_K_L.gguf) | Q4_K_L | 5.84GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. | +| [granite-4.1-8b-Q4_1.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q4_1.gguf) | Q4_1 | 5.65GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. | +| [granite-4.1-8b-Q4_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q4_K_M.gguf) | Q4_K_M | 5.54GB | false | Good quality, default size for most use cases, *recommended*. | +| [granite-4.1-8b-Q4_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q4_K_S.gguf) | Q4_K_S | 5.19GB | false | Slightly lower quality with more space savings, *recommended*. | +| [granite-4.1-8b-IQ4_NL.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ4_NL.gguf) | IQ4_NL | 5.19GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. | +| [granite-4.1-8b-Q4_0.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q4_0.gguf) | Q4_0 | 5.16GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. | +| [granite-4.1-8b-Q3_K_XL.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q3_K_XL.gguf) | Q3_K_XL | 5.11GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. | +| [granite-4.1-8b-IQ4_XS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ4_XS.gguf) | IQ4_XS | 4.94GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. | +| [granite-4.1-8b-Q3_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q3_K_L.gguf) | Q3_K_L | 4.75GB | false | Lower quality but usable, good for low RAM availability. | +| [granite-4.1-8b-Q3_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q3_K_M.gguf) | Q3_K_M | 4.49GB | false | Low quality. | +| [granite-4.1-8b-IQ3_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ3_M.gguf) | IQ3_M | 4.24GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. | +| [granite-4.1-8b-Q3_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q3_K_S.gguf) | Q3_K_S | 4.07GB | false | Low quality, not recommended. | +| [granite-4.1-8b-Q2_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q2_K_L.gguf) | Q2_K_L | 3.98GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. | +| [granite-4.1-8b-IQ3_XS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ3_XS.gguf) | IQ3_XS | 3.95GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. | +| [granite-4.1-8b-IQ3_XXS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ3_XXS.gguf) | IQ3_XXS | 3.72GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. | +| [granite-4.1-8b-Q2_K.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-Q2_K.gguf) | Q2_K | 3.58GB | false | Very low quality but surprisingly usable. | +| [granite-4.1-8b-IQ2_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-4.1-8b-GGUF/blob/main/ibm-granite_granite-4.1-8b-IQ2_M.gguf) | IQ2_M | 3.51GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. | + +## 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. + +## Downloading using huggingface-cli + +
+ Click to view download instructions + +First, make sure you have hugginface-cli installed: + +``` +pip install -U "huggingface_hub[cli]" +``` + +Then, you can target the specific file you want: + +``` +huggingface-cli download bartowski/ibm-granite_granite-4.1-8b-GGUF --include "ibm-granite_granite-4.1-8b-Q4_K_M.gguf" --local-dir ./ +``` + +If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run: + +``` +huggingface-cli download bartowski/ibm-granite_granite-4.1-8b-GGUF --include "ibm-granite_granite-4.1-8b-Q8_0/*" --local-dir ./ +``` + +You can either specify a new local-dir (ibm-granite_granite-4.1-8b-Q8_0) or download them all in place (./) + +
+ +## ARM/AVX information + +Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass. + +Now, however, there is something called "online repacking" for weights. details in [this PR](https://github.com/ggml-org/llama.cpp/pull/9921). If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly. + +As of llama.cpp build [b4282](https://github.com/ggml-org/llama.cpp/releases/tag/b4282) you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0. + +Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to [this PR](https://github.com/ggml-org/llama.cpp/pull/10541) which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase. + +
+ Click to view Q4_0_X_X information (deprecated + +I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking. + +
+ Click to view benchmarks on an AVX2 system (EPYC7702) + +| model | size | params | backend | threads | test | t/s | % (vs Q4_0) | +| ------------------------------ | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |-------------: | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp512 | 204.03 ± 1.03 | 100% | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp1024 | 282.92 ± 0.19 | 100% | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp2048 | 259.49 ± 0.44 | 100% | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg128 | 39.12 ± 0.27 | 100% | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg256 | 39.31 ± 0.69 | 100% | +| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg512 | 40.52 ± 0.03 | 100% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp512 | 301.02 ± 1.74 | 147% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp1024 | 287.23 ± 0.20 | 101% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp2048 | 262.77 ± 1.81 | 101% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg128 | 18.80 ± 0.99 | 48% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg256 | 24.46 ± 3.04 | 83% | +| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg512 | 36.32 ± 3.59 | 90% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp512 | 271.71 ± 3.53 | 133% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp1024 | 279.86 ± 45.63 | 100% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp2048 | 320.77 ± 5.00 | 124% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg128 | 43.51 ± 0.05 | 111% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg256 | 43.35 ± 0.09 | 110% | +| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg512 | 42.60 ± 0.31 | 105% | + +Q4_0_8_8 offers a nice bump to prompt processing and a small bump to text generation + +
+ +
+ +## Which file should I choose? + +
+ Click here for details + +A great write up with charts showing various performances is provided by Artefact2 [here](https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9) + +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. + +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](https://github.com/ggml-org/llama.cpp/wiki/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. + +Thank you to LM Studio for sponsoring my work. + +Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski diff --git a/configuration.json b/configuration.json new file mode 100644 index 0000000..bbeeda1 --- /dev/null +++ b/configuration.json @@ -0,0 +1 @@ +{"framework": "pytorch", "task": "text-generation", "allow_remote": true} \ No newline at end of file diff --git a/ibm-granite_granite-4.1-8b-IQ2_M.gguf b/ibm-granite_granite-4.1-8b-IQ2_M.gguf new file mode 100644 index 0000000..260f651 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ2_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04623b297d113eb8e83bba39e9d9642bee446e17ac6f541cb38ade1d88f1ba91 +size 3510383648 diff --git a/ibm-granite_granite-4.1-8b-IQ3_M.gguf b/ibm-granite_granite-4.1-8b-IQ3_M.gguf new file mode 100644 index 0000000..aa1133f --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ3_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8133437d4256da3cda37d55011effeafd7b5d543e4f80335b8fcfadebe4b0438 +size 4241224736 diff --git a/ibm-granite_granite-4.1-8b-IQ3_XS.gguf b/ibm-granite_granite-4.1-8b-IQ3_XS.gguf new file mode 100644 index 0000000..4446833 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ3_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17dea092994f4aa74543d4df147aabde3e4c1beb7a97836ca392db39dfccc0d2 +size 3948852256 diff --git a/ibm-granite_granite-4.1-8b-IQ3_XXS.gguf b/ibm-granite_granite-4.1-8b-IQ3_XXS.gguf new file mode 100644 index 0000000..163c4d2 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ3_XXS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:711813ebfe689246a3f9e51d088ff877cfe247bb58bd41a406316f7f7521c76f +size 3723211808 diff --git a/ibm-granite_granite-4.1-8b-IQ4_NL.gguf b/ibm-granite_granite-4.1-8b-IQ4_NL.gguf new file mode 100644 index 0000000..8d8925d --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ4_NL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20174f1dda23e9831a168da52cb641c442a8acf757981588fbb13bd15fae8af3 +size 5185713184 diff --git a/ibm-granite_granite-4.1-8b-IQ4_XS.gguf b/ibm-granite_granite-4.1-8b-IQ4_XS.gguf new file mode 100644 index 0000000..6d4b604 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-IQ4_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86cb2036d3a4d40295345b92a7a35423b6075278152561e659b6bcf54128877d +size 4939559968 diff --git a/ibm-granite_granite-4.1-8b-Q2_K.gguf b/ibm-granite_granite-4.1-8b-Q2_K.gguf new file mode 100644 index 0000000..813d665 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q2_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fca0ccf472571774b8973c8670bc43b638eb3baf91999088758ce64b628957ed +size 3577459744 diff --git a/ibm-granite_granite-4.1-8b-Q2_K_L.gguf b/ibm-granite_granite-4.1-8b-Q2_K_L.gguf new file mode 100644 index 0000000..02d9284 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q2_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0feb210214c58694df9a7498b1da283dda4f4055fcbc2ed69695dab11c4c0a9 +size 3978867744 diff --git a/ibm-granite_granite-4.1-8b-Q3_K_L.gguf b/ibm-granite_granite-4.1-8b-Q3_K_L.gguf new file mode 100644 index 0000000..20ddd0a --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q3_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22d0cf1282d8a1cad8b05a1b45ff04374c79c03b3ac2e0c00abb97388ad8a288 +size 4747080736 diff --git a/ibm-granite_granite-4.1-8b-Q3_K_M.gguf b/ibm-granite_granite-4.1-8b-Q3_K_M.gguf new file mode 100644 index 0000000..57a23a9 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q3_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b91c4253fe63f14a0f5643e5b1772541da346f22dce25263a0d98a8d10f4c45 +size 4487558176 diff --git a/ibm-granite_granite-4.1-8b-Q3_K_S.gguf b/ibm-granite_granite-4.1-8b-Q3_K_S.gguf new file mode 100644 index 0000000..ed53d6d --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q3_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc6673597c9ee3e2328ec3a4c86f408c3f5fb0be3cb6b82666a46ac423e497b6 +size 4074026016 diff --git a/ibm-granite_granite-4.1-8b-Q3_K_XL.gguf b/ibm-granite_granite-4.1-8b-Q3_K_XL.gguf new file mode 100644 index 0000000..dce3a18 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q3_K_XL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79c28fbf546432bcee980a38afb77a1eea418a8430dcf7d9923ceec0ac14cef3 +size 5106742304 diff --git a/ibm-granite_granite-4.1-8b-Q4_0.gguf b/ibm-granite_granite-4.1-8b-Q4_0.gguf new file mode 100644 index 0000000..765b30f --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q4_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4ff64020a768c52f2e595785c5f262121d5b8a42b91ef640772a0618a70e3b7 +size 5158843424 diff --git a/ibm-granite_granite-4.1-8b-Q4_1.gguf b/ibm-granite_granite-4.1-8b-Q4_1.gguf new file mode 100644 index 0000000..00e9a14 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q4_1.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42399be4cba3ec54187801fb12d969937260f141c66b45704adb96df233aa3e0 +size 5645251616 diff --git a/ibm-granite_granite-4.1-8b-Q4_K_L.gguf b/ibm-granite_granite-4.1-8b-Q4_K_L.gguf new file mode 100644 index 0000000..c36d749 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q4_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d01b100b91ba76dfc22fdd473cd1f873d7993633e72d68f01bfce0c3c0916ff +size 5844349984 diff --git a/ibm-granite_granite-4.1-8b-Q4_K_M.gguf b/ibm-granite_granite-4.1-8b-Q4_K_M.gguf new file mode 100644 index 0000000..1d8e2aa --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q4_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f42811b446eeddf8ca50a287e0c49abafb11f0ca8ed23606fd758720d2a351f +size 5539279904 diff --git a/ibm-granite_granite-4.1-8b-Q4_K_S.gguf b/ibm-granite_granite-4.1-8b-Q4_K_S.gguf new file mode 100644 index 0000000..661eb75 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q4_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea58dbc71629e28bae4391eabe18903fd210643e4e5630e970c1380e14c4a2b9 +size 5185713184 diff --git a/ibm-granite_granite-4.1-8b-Q5_K_L.gguf b/ibm-granite_granite-4.1-8b-Q5_K_L.gguf new file mode 100644 index 0000000..2211401 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q5_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e400eaee7dba772d21d1d9f5b8d688f205158c670fec8bd84de99623d3d20847 +size 6663549984 diff --git a/ibm-granite_granite-4.1-8b-Q5_K_M.gguf b/ibm-granite_granite-4.1-8b-Q5_K_M.gguf new file mode 100644 index 0000000..7cb3bab --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q5_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5ec471bd46234c4b3064794551b0cc0f76aab264157454ebfd055cf03e36bb8 +size 6409860128 diff --git a/ibm-granite_granite-4.1-8b-Q5_K_S.gguf b/ibm-granite_granite-4.1-8b-Q5_K_S.gguf new file mode 100644 index 0000000..1b6ff3e --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q5_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9cb5079cc3e75aadf6c9030d714789fd72ee128876c705bdbd48f9602d03b9b3 +size 6148043808 diff --git a/ibm-granite_granite-4.1-8b-Q6_K.gguf b/ibm-granite_granite-4.1-8b-Q6_K.gguf new file mode 100644 index 0000000..2f8d94e --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q6_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffe411fc631ee05f6a390d696c175374808ab8bc920a9ac6ba933bfde490c1e7 +size 7521219616 diff --git a/ibm-granite_granite-4.1-8b-Q6_K_L.gguf b/ibm-granite_granite-4.1-8b-Q6_K_L.gguf new file mode 100644 index 0000000..e4a7d10 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q6_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ac6d0f667a596c9bcc4529845c05c6451c6d8f25b39faf35d5802850dc4f959 +size 7720317984 diff --git a/ibm-granite_granite-4.1-8b-Q8_0.gguf b/ibm-granite_granite-4.1-8b-Q8_0.gguf new file mode 100644 index 0000000..a0e5ca6 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-Q8_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82b5c4585b1bf357a5f46eea9906f8599e0c65d3c5fadcae80ca6c19104796de +size 9345610784 diff --git a/ibm-granite_granite-4.1-8b-bf16.gguf b/ibm-granite_granite-4.1-8b-bf16.gguf new file mode 100644 index 0000000..aa24b37 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-bf16.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c426fa9bcad2ec67f39fb3b1e413c89aec28020e9feeca68aabc81a308cea5c7 +size 17587417856 diff --git a/ibm-granite_granite-4.1-8b-imatrix.gguf b/ibm-granite_granite-4.1-8b-imatrix.gguf new file mode 100644 index 0000000..e9c6fb8 --- /dev/null +++ b/ibm-granite_granite-4.1-8b-imatrix.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e638650c621ce10aac2c9a15c3d78d299a8ff7fd4637733dc0782ddc4feaba7e +size 6023264