From 7a3e573ffaf0fd6d6afc96c0509507fbfb3232df Mon Sep 17 00:00:00 2001 From: ModelHub XC Date: Wed, 15 Jul 2026 05:21:05 +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/THUDM_GLM-4-9B-0414-GGUF Source: Original Platform --- .gitattributes | 72 +++++++++++++ README.md | 176 +++++++++++++++++++++++++++++++ THUDM_GLM-4-9B-0414-IQ2_M.gguf | 3 + THUDM_GLM-4-9B-0414-IQ3_M.gguf | 3 + THUDM_GLM-4-9B-0414-IQ3_XS.gguf | 3 + THUDM_GLM-4-9B-0414-IQ3_XXS.gguf | 3 + THUDM_GLM-4-9B-0414-IQ4_NL.gguf | 3 + THUDM_GLM-4-9B-0414-IQ4_XS.gguf | 3 + THUDM_GLM-4-9B-0414-Q2_K.gguf | 3 + THUDM_GLM-4-9B-0414-Q2_K_L.gguf | 3 + THUDM_GLM-4-9B-0414-Q3_K_L.gguf | 3 + THUDM_GLM-4-9B-0414-Q3_K_M.gguf | 3 + THUDM_GLM-4-9B-0414-Q3_K_S.gguf | 3 + THUDM_GLM-4-9B-0414-Q3_K_XL.gguf | 3 + THUDM_GLM-4-9B-0414-Q4_0.gguf | 3 + THUDM_GLM-4-9B-0414-Q4_1.gguf | 3 + THUDM_GLM-4-9B-0414-Q4_K_L.gguf | 3 + THUDM_GLM-4-9B-0414-Q4_K_M.gguf | 3 + THUDM_GLM-4-9B-0414-Q4_K_S.gguf | 3 + THUDM_GLM-4-9B-0414-Q5_K_L.gguf | 3 + THUDM_GLM-4-9B-0414-Q5_K_M.gguf | 3 + THUDM_GLM-4-9B-0414-Q5_K_S.gguf | 3 + THUDM_GLM-4-9B-0414-Q6_K.gguf | 3 + THUDM_GLM-4-9B-0414-Q6_K_L.gguf | 3 + THUDM_GLM-4-9B-0414-Q8_0.gguf | 3 + THUDM_GLM-4-9B-0414-bf16.gguf | 3 + THUDM_GLM-4-9B-0414.imatrix | 3 + configuration.json | 1 + 28 files changed, 324 insertions(+) create mode 100644 .gitattributes create mode 100644 README.md create mode 100644 THUDM_GLM-4-9B-0414-IQ2_M.gguf create mode 100644 THUDM_GLM-4-9B-0414-IQ3_M.gguf create mode 100644 THUDM_GLM-4-9B-0414-IQ3_XS.gguf create mode 100644 THUDM_GLM-4-9B-0414-IQ3_XXS.gguf create mode 100644 THUDM_GLM-4-9B-0414-IQ4_NL.gguf create mode 100644 THUDM_GLM-4-9B-0414-IQ4_XS.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q2_K.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q2_K_L.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q3_K_L.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q3_K_M.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q3_K_S.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q3_K_XL.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q4_0.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q4_1.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q4_K_L.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q4_K_M.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q4_K_S.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q5_K_L.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q5_K_M.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q5_K_S.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q6_K.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q6_K_L.gguf create mode 100644 THUDM_GLM-4-9B-0414-Q8_0.gguf create mode 100644 THUDM_GLM-4-9B-0414-bf16.gguf create mode 100644 THUDM_GLM-4-9B-0414.imatrix create mode 100644 configuration.json diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..a347331 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,72 @@ +*.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 +*.gguf* 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 +THUDM_GLM-4-9B-0414.imatrix filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ3_XS.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ4_NL.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q4_1.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q5_K_L.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q6_K_L.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text +THUDM_GLM-4-9B-0414-bf16.gguf filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..04f0fd0 --- /dev/null +++ b/README.md @@ -0,0 +1,176 @@ +--- +quantized_by: bartowski +pipeline_tag: text-generation +language: +- zh +- en +license: mit +base_model: THUDM/GLM-4-9B-0414 +base_model_relation: quantized +--- + +## Llamacpp imatrix Quantizations of GLM-4-9B-0414 by THUDM + +Using llama.cpp release b5228 for quantization. + +Original model: https://huggingface.co/THUDM/GLM-4-9B-0414 + +All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8) + +Run them in [LM Studio](https://lmstudio.ai/) + +Run them directly with [llama.cpp](https://github.com/ggerganov/llama.cpp), or any other llama.cpp based project + +## Prompt format + +``` +[gMASK]<|system|> +{system_prompt}<|user|> +{prompt}<|assistant|> +``` + +## What's new: + +Fix tokenizer + +## Download a file (not the whole branch) from below: + +| Filename | Quant type | File Size | Split | Description | +| -------- | ---------- | --------- | ----- | ----------- | +| [GLM-4-9B-0414-bf16.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-bf16.gguf) | bf16 | 18.81GB | false | Full BF16 weights. | +| [GLM-4-9B-0414-Q8_0.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q8_0.gguf) | Q8_0 | 10.00GB | false | Extremely high quality, generally unneeded but max available quant. | +| [GLM-4-9B-0414-Q6_K_L.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q6_K_L.gguf) | Q6_K_L | 8.57GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. | +| [GLM-4-9B-0414-Q6_K.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q6_K.gguf) | Q6_K | 8.27GB | false | Very high quality, near perfect, *recommended*. | +| [GLM-4-9B-0414-Q5_K_L.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q5_K_L.gguf) | Q5_K_L | 7.43GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. | +| [GLM-4-9B-0414-Q5_K_M.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q5_K_M.gguf) | Q5_K_M | 7.05GB | false | High quality, *recommended*. | +| [GLM-4-9B-0414-Q5_K_S.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q5_K_S.gguf) | Q5_K_S | 6.70GB | false | High quality, *recommended*. | +| [GLM-4-9B-0414-Q4_K_L.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q4_K_L.gguf) | Q4_K_L | 6.63GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. | +| [GLM-4-9B-0414-Q4_K_M.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q4_K_M.gguf) | Q4_K_M | 6.17GB | false | Good quality, default size for most use cases, *recommended*. | +| [GLM-4-9B-0414-Q4_1.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q4_1.gguf) | Q4_1 | 6.01GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. | +| [GLM-4-9B-0414-Q4_K_S.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q4_K_S.gguf) | Q4_K_S | 5.76GB | false | Slightly lower quality with more space savings, *recommended*. | +| [GLM-4-9B-0414-Q3_K_XL.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q3_K_XL.gguf) | Q3_K_XL | 5.74GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. | +| [GLM-4-9B-0414-Q4_0.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q4_0.gguf) | Q4_0 | 5.48GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. | +| [GLM-4-9B-0414-IQ4_NL.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ4_NL.gguf) | IQ4_NL | 5.47GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. | +| [GLM-4-9B-0414-IQ4_XS.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ4_XS.gguf) | IQ4_XS | 5.26GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. | +| [GLM-4-9B-0414-Q3_K_L.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q3_K_L.gguf) | Q3_K_L | 5.20GB | false | Lower quality but usable, good for low RAM availability. | +| [GLM-4-9B-0414-Q3_K_M.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q3_K_M.gguf) | Q3_K_M | 4.97GB | false | Low quality. | +| [GLM-4-9B-0414-IQ3_M.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ3_M.gguf) | IQ3_M | 4.72GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. | +| [GLM-4-9B-0414-Q2_K_L.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q2_K_L.gguf) | Q2_K_L | 4.61GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. | +| [GLM-4-9B-0414-Q3_K_S.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q3_K_S.gguf) | Q3_K_S | 4.59GB | false | Low quality, not recommended. | +| [GLM-4-9B-0414-IQ3_XS.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ3_XS.gguf) | IQ3_XS | 4.41GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. | +| [GLM-4-9B-0414-IQ3_XXS.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ3_XXS.gguf) | IQ3_XXS | 4.23GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. | +| [GLM-4-9B-0414-Q2_K.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-Q2_K.gguf) | Q2_K | 4.01GB | false | Very low quality but surprisingly usable. | +| [GLM-4-9B-0414-IQ2_M.gguf](https://huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF/blob/main/THUDM_GLM-4-9B-0414-IQ2_M.gguf) | IQ2_M | 3.95GB | 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/THUDM_GLM-4-9B-0414-GGUF --include "THUDM_GLM-4-9B-0414-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/THUDM_GLM-4-9B-0414-GGUF --include "THUDM_GLM-4-9B-0414-Q8_0/*" --local-dir ./ +``` + +You can either specify a new local-dir (THUDM_GLM-4-9B-0414-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/ggerganov/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/ggerganov/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/ggerganov/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/ggerganov/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/THUDM_GLM-4-9B-0414-IQ2_M.gguf b/THUDM_GLM-4-9B-0414-IQ2_M.gguf new file mode 100644 index 0000000..0e61ecd --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ2_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2b96ccbb2f674f60b672191f4beeb68a433264dec38b6124689030417058863 +size 3946313088 diff --git a/THUDM_GLM-4-9B-0414-IQ3_M.gguf b/THUDM_GLM-4-9B-0414-IQ3_M.gguf new file mode 100644 index 0000000..fcca9f0 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ3_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3473c34a3d95c4e1e471896850b1243d01dc0d0d77b568363b84bfa8fb8f6dec +size 4721800576 diff --git a/THUDM_GLM-4-9B-0414-IQ3_XS.gguf b/THUDM_GLM-4-9B-0414-IQ3_XS.gguf new file mode 100644 index 0000000..4e944c8 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ3_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d619395c61df952f527dc8c956e56e18db21e254789a0ca16df07b2ddb45b220 +size 4406408576 diff --git a/THUDM_GLM-4-9B-0414-IQ3_XXS.gguf b/THUDM_GLM-4-9B-0414-IQ3_XXS.gguf new file mode 100644 index 0000000..abd04dd --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ3_XXS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82909b3b3673f8750599db35e45829a14f4ff206384004489afe7c37d072866c +size 4226807168 diff --git a/THUDM_GLM-4-9B-0414-IQ4_NL.gguf b/THUDM_GLM-4-9B-0414-IQ4_NL.gguf new file mode 100644 index 0000000..05a6485 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ4_NL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea23ab9fe9758715a42c5f2c99f1c27ac3c720146bbadef5c3599df970442cd2 +size 5465175424 diff --git a/THUDM_GLM-4-9B-0414-IQ4_XS.gguf b/THUDM_GLM-4-9B-0414-IQ4_XS.gguf new file mode 100644 index 0000000..a18cce3 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-IQ4_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c85b661ed11c36f8b5a4f75da8bc1672b011febf8d71f098feab41186ad1767e +size 5262275968 diff --git a/THUDM_GLM-4-9B-0414-Q2_K.gguf b/THUDM_GLM-4-9B-0414-Q2_K.gguf new file mode 100644 index 0000000..e97f6f9 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q2_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f6ad4ae6d9bb4cca4de34694211bc669bf8df1e71b69c4c2629fc8608ecf1bf +size 4006344064 diff --git a/THUDM_GLM-4-9B-0414-Q2_K_L.gguf b/THUDM_GLM-4-9B-0414-Q2_K_L.gguf new file mode 100644 index 0000000..355fa5c --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q2_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:487cab46b3ecaaeb95b1dc9e085dec5d6d5dbbbda2964a68af77cc41132637fc +size 4612552064 diff --git a/THUDM_GLM-4-9B-0414-Q3_K_L.gguf b/THUDM_GLM-4-9B-0414-Q3_K_L.gguf new file mode 100644 index 0000000..8bd287e --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q3_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36c5dd25a85660d040d6d28a4398fc636d5d59521bcab3552e05720f99a0b8a7 +size 5196608896 diff --git a/THUDM_GLM-4-9B-0414-Q3_K_M.gguf b/THUDM_GLM-4-9B-0414-Q3_K_M.gguf new file mode 100644 index 0000000..ad6dfb2 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q3_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bac1470340530679564b3c3b670494387f996004603a3af619c484fb18515f55 +size 4974507392 diff --git a/THUDM_GLM-4-9B-0414-Q3_K_S.gguf b/THUDM_GLM-4-9B-0414-Q3_K_S.gguf new file mode 100644 index 0000000..ae731cf --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q3_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2385616559e8ed2314c501dc5f6a85f7c8348d02ee38ef4a4ff592eda02160c +size 4592039296 diff --git a/THUDM_GLM-4-9B-0414-Q3_K_XL.gguf b/THUDM_GLM-4-9B-0414-Q3_K_XL.gguf new file mode 100644 index 0000000..1167ba8 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q3_K_XL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f0d41f07bc51aeb0453a034a56a74c7963ef5343b65b4e4d6d8bbbe3d0025a6 +size 5739771264 diff --git a/THUDM_GLM-4-9B-0414-Q4_0.gguf b/THUDM_GLM-4-9B-0414-Q4_0.gguf new file mode 100644 index 0000000..192efe4 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q4_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4996db1053041e801ca57e030a1ea3ad70b6d3eeb49573092313bd7776eecdf5 +size 5477463424 diff --git a/THUDM_GLM-4-9B-0414-Q4_1.gguf b/THUDM_GLM-4-9B-0414-Q4_1.gguf new file mode 100644 index 0000000..14b8d3d --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q4_1.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:507ffe123f0e561a2f9c5dde84b3d23d8d9332a7e244e2b5cfb28e02eb81cc83 +size 6008599936 diff --git a/THUDM_GLM-4-9B-0414-Q4_K_L.gguf b/THUDM_GLM-4-9B-0414-Q4_K_L.gguf new file mode 100644 index 0000000..7f99123 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q4_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f631f071e4e91789bbf362affbe0fd217625dd3054dec7aa078a2c7f2ff3da5f +size 6627292544 diff --git a/THUDM_GLM-4-9B-0414-Q4_K_M.gguf b/THUDM_GLM-4-9B-0414-Q4_K_M.gguf new file mode 100644 index 0000000..90b58f5 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q4_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b95f0b88eca423bfb874c4f8e7b9574ec76acab2befd272076e8a8dce31216a +size 6166574464 diff --git a/THUDM_GLM-4-9B-0414-Q4_K_S.gguf b/THUDM_GLM-4-9B-0414-Q4_K_S.gguf new file mode 100644 index 0000000..661668f --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q4_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edecb34f065c51223826618b35bb47da1d13f4d2fbddb3a99c51db46cb22bf95 +size 5758481792 diff --git a/THUDM_GLM-4-9B-0414-Q5_K_L.gguf b/THUDM_GLM-4-9B-0414-Q5_K_L.gguf new file mode 100644 index 0000000..f472192 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q5_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c43f4d4949504e1a13617269d40f3cb71a82e1c4d47817e372cf08704fbd7a9c +size 7434040704 diff --git a/THUDM_GLM-4-9B-0414-Q5_K_M.gguf b/THUDM_GLM-4-9B-0414-Q5_K_M.gguf new file mode 100644 index 0000000..556b21d --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q5_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24013ada7a32e4e449e8aad7afb48261abb1c1d86be55506065c3676d0ad6742 +size 7050917248 diff --git a/THUDM_GLM-4-9B-0414-Q5_K_S.gguf b/THUDM_GLM-4-9B-0414-Q5_K_S.gguf new file mode 100644 index 0000000..128db19 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q5_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29f7419db2f8e1e6ccb102aa5400f67c2d7c1553a7270705074b6c9f9eb3ee94 +size 6697514368 diff --git a/THUDM_GLM-4-9B-0414-Q6_K.gguf b/THUDM_GLM-4-9B-0414-Q6_K.gguf new file mode 100644 index 0000000..221049b --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q6_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b20636a8389d30295959b64df48e3e4a0de29ea00a6abb9be494bf1773858b7c +size 8266642816 diff --git a/THUDM_GLM-4-9B-0414-Q6_K_L.gguf b/THUDM_GLM-4-9B-0414-Q6_K_L.gguf new file mode 100644 index 0000000..37a8f0e --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q6_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79c3818e023caa6e445f17eac85b741d4ce648cc5e4096dfdb8eef07a298c8c0 +size 8567321984 diff --git a/THUDM_GLM-4-9B-0414-Q8_0.gguf b/THUDM_GLM-4-9B-0414-Q8_0.gguf new file mode 100644 index 0000000..020a43d --- /dev/null +++ b/THUDM_GLM-4-9B-0414-Q8_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b4ea2795934ca05dc409251dddd289160a194a4d920b575d1de4516808bb50d +size 9999611264 diff --git a/THUDM_GLM-4-9B-0414-bf16.gguf b/THUDM_GLM-4-9B-0414-bf16.gguf new file mode 100644 index 0000000..e981c83 --- /dev/null +++ b/THUDM_GLM-4-9B-0414-bf16.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:375fd89fc82886616ece3cdcdce9364983f54059cbb4ab1e027c1a171d5f1769 +size 18811581568 diff --git a/THUDM_GLM-4-9B-0414.imatrix b/THUDM_GLM-4-9B-0414.imatrix new file mode 100644 index 0000000..c88a0b2 --- /dev/null +++ b/THUDM_GLM-4-9B-0414.imatrix @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6797698787046c9bb5ff6414fe58961e8dedb74da18e84fec02f1a84dd413410 +size 5476108 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