From 1a4acaab45ea7088f426a53f93bd686860d4d5a1 Mon Sep 17 00:00:00 2001 From: ModelHub XC Date: Sat, 1 Aug 2026 21:45:13 +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/Qwen_Qwen3-8B-GGUF Source: Original Platform --- .gitattributes | 49 +++++++++++ Qwen_Qwen3-8B-IQ2_M.gguf | 3 + Qwen_Qwen3-8B-IQ3_M.gguf | 3 + Qwen_Qwen3-8B-IQ3_XS.gguf | 3 + Qwen_Qwen3-8B-IQ3_XXS.gguf | 3 + Qwen_Qwen3-8B-IQ4_NL.gguf | 3 + Qwen_Qwen3-8B-IQ4_XS.gguf | 3 + Qwen_Qwen3-8B-Q2_K.gguf | 3 + Qwen_Qwen3-8B-Q2_K_L.gguf | 3 + Qwen_Qwen3-8B-Q3_K_L.gguf | 3 + Qwen_Qwen3-8B-Q3_K_M.gguf | 3 + Qwen_Qwen3-8B-Q3_K_S.gguf | 3 + Qwen_Qwen3-8B-Q3_K_XL.gguf | 3 + Qwen_Qwen3-8B-Q4_0.gguf | 3 + Qwen_Qwen3-8B-Q4_1.gguf | 3 + Qwen_Qwen3-8B-Q4_K_L.gguf | 3 + Qwen_Qwen3-8B-Q4_K_M.gguf | 3 + Qwen_Qwen3-8B-Q4_K_S.gguf | 3 + Qwen_Qwen3-8B-Q5_K_L.gguf | 3 + Qwen_Qwen3-8B-Q5_K_M.gguf | 3 + Qwen_Qwen3-8B-Q5_K_S.gguf | 3 + Qwen_Qwen3-8B-Q6_K.gguf | 3 + Qwen_Qwen3-8B-Q6_K_L.gguf | 3 + Qwen_Qwen3-8B-Q8_0.gguf | 3 + Qwen_Qwen3-8B-bf16.gguf | 3 + Qwen_Qwen3-8B.imatrix | 3 + README.md | 171 +++++++++++++++++++++++++++++++++++++ configuration.json | 1 + 28 files changed, 296 insertions(+) create mode 100644 .gitattributes create mode 100644 Qwen_Qwen3-8B-IQ2_M.gguf create mode 100644 Qwen_Qwen3-8B-IQ3_M.gguf create mode 100644 Qwen_Qwen3-8B-IQ3_XS.gguf create mode 100644 Qwen_Qwen3-8B-IQ3_XXS.gguf create mode 100644 Qwen_Qwen3-8B-IQ4_NL.gguf create mode 100644 Qwen_Qwen3-8B-IQ4_XS.gguf create mode 100644 Qwen_Qwen3-8B-Q2_K.gguf create mode 100644 Qwen_Qwen3-8B-Q2_K_L.gguf create mode 100644 Qwen_Qwen3-8B-Q3_K_L.gguf create mode 100644 Qwen_Qwen3-8B-Q3_K_M.gguf create mode 100644 Qwen_Qwen3-8B-Q3_K_S.gguf create mode 100644 Qwen_Qwen3-8B-Q3_K_XL.gguf create mode 100644 Qwen_Qwen3-8B-Q4_0.gguf create mode 100644 Qwen_Qwen3-8B-Q4_1.gguf create mode 100644 Qwen_Qwen3-8B-Q4_K_L.gguf create mode 100644 Qwen_Qwen3-8B-Q4_K_M.gguf create mode 100644 Qwen_Qwen3-8B-Q4_K_S.gguf create mode 100644 Qwen_Qwen3-8B-Q5_K_L.gguf create mode 100644 Qwen_Qwen3-8B-Q5_K_M.gguf create mode 100644 Qwen_Qwen3-8B-Q5_K_S.gguf create mode 100644 Qwen_Qwen3-8B-Q6_K.gguf create mode 100644 Qwen_Qwen3-8B-Q6_K_L.gguf create mode 100644 Qwen_Qwen3-8B-Q8_0.gguf create mode 100644 Qwen_Qwen3-8B-bf16.gguf create mode 100644 Qwen_Qwen3-8B.imatrix create mode 100644 README.md create mode 100644 configuration.json diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..9383a92 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,49 @@ +*.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 + +Qwen_Qwen3-8B.imatrix filter=lfs diff=lfs merge=lfs -text \ No newline at end of file diff --git a/Qwen_Qwen3-8B-IQ2_M.gguf b/Qwen_Qwen3-8B-IQ2_M.gguf new file mode 100644 index 0000000..929eb3b --- /dev/null +++ b/Qwen_Qwen3-8B-IQ2_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38833d850e00f75bac26c0c848f2fde9100a5a21c18aae89cfd8611efca6194b +size 3051914784 diff --git a/Qwen_Qwen3-8B-IQ3_M.gguf b/Qwen_Qwen3-8B-IQ3_M.gguf new file mode 100644 index 0000000..fd4a38c --- /dev/null +++ b/Qwen_Qwen3-8B-IQ3_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de1982c971c26a256dc2ac6d04b04f1cd20ea35e0d3ec3263320756dea66bee3 +size 3896620576 diff --git a/Qwen_Qwen3-8B-IQ3_XS.gguf b/Qwen_Qwen3-8B-IQ3_XS.gguf new file mode 100644 index 0000000..0c4335d --- /dev/null +++ b/Qwen_Qwen3-8B-IQ3_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15d9f1cfa8be0785df44376669a5b45c3ae75eda916fc553c44923cfb5c3eef2 +size 3626874400 diff --git a/Qwen_Qwen3-8B-IQ3_XXS.gguf b/Qwen_Qwen3-8B-IQ3_XXS.gguf new file mode 100644 index 0000000..b4f6ceb --- /dev/null +++ b/Qwen_Qwen3-8B-IQ3_XXS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11a2ba2a7970f621c03e05bce111df2cb7c04880ffec2cdd03445d7b5e8b8514 +size 3369633312 diff --git a/Qwen_Qwen3-8B-IQ4_NL.gguf b/Qwen_Qwen3-8B-IQ4_NL.gguf new file mode 100644 index 0000000..16ffb18 --- /dev/null +++ b/Qwen_Qwen3-8B-IQ4_NL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57d2f29c36469e05cf7801704b92d42dfc24531a9f19e92b2a2b4d0c4d9b7545 +size 4793624096 diff --git a/Qwen_Qwen3-8B-IQ4_XS.gguf b/Qwen_Qwen3-8B-IQ4_XS.gguf new file mode 100644 index 0000000..7808d12 --- /dev/null +++ b/Qwen_Qwen3-8B-IQ4_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f69fe02ca9764c93bcc12ac96509bee53488114d6a7fe47764f80990c28751a +size 4561839648 diff --git a/Qwen_Qwen3-8B-Q2_K.gguf b/Qwen_Qwen3-8B-Q2_K.gguf new file mode 100644 index 0000000..18a47bb --- /dev/null +++ b/Qwen_Qwen3-8B-Q2_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e89189c258704b98b6b0b5bdc1f1b168f523a706b2d9d3e9b5aefbf94ac5de32 +size 3281733152 diff --git a/Qwen_Qwen3-8B-Q2_K_L.gguf b/Qwen_Qwen3-8B-Q2_K_L.gguf new file mode 100644 index 0000000..68b6f33 --- /dev/null +++ b/Qwen_Qwen3-8B-Q2_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79f2c07dc0d7be0c6db8192beb670cff82d173c38c05f5a9ec4c0104e26736fd +size 3889477152 diff --git a/Qwen_Qwen3-8B-Q3_K_L.gguf b/Qwen_Qwen3-8B-Q3_K_L.gguf new file mode 100644 index 0000000..035c5b1 --- /dev/null +++ b/Qwen_Qwen3-8B-Q3_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d34d3a93be237268df8437109a7e461a9c43136106d391fba21d7b3db7794a4d +size 4431394336 diff --git a/Qwen_Qwen3-8B-Q3_K_M.gguf b/Qwen_Qwen3-8B-Q3_K_M.gguf new file mode 100644 index 0000000..d70706f --- /dev/null +++ b/Qwen_Qwen3-8B-Q3_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2c61b55d39ec6fc43f84b4cbbed4505fde386d4e03ed005087bfd8c69c502c3 +size 4124161568 diff --git a/Qwen_Qwen3-8B-Q3_K_S.gguf b/Qwen_Qwen3-8B-Q3_K_S.gguf new file mode 100644 index 0000000..9ba3d02 --- /dev/null +++ b/Qwen_Qwen3-8B-Q3_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d03b4542b0b11ac88fceaac135d946c528e3f0469384926f6e94755315174410 +size 3769611808 diff --git a/Qwen_Qwen3-8B-Q3_K_XL.gguf b/Qwen_Qwen3-8B-Q3_K_XL.gguf new file mode 100644 index 0000000..e57fd56 --- /dev/null +++ b/Qwen_Qwen3-8B-Q3_K_XL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b0454730e97c9567bc9f60d1ca5125a5e99750b05f81978744d76f3416d74bc +size 4975932960 diff --git a/Qwen_Qwen3-8B-Q4_0.gguf b/Qwen_Qwen3-8B-Q4_0.gguf new file mode 100644 index 0000000..6e5a8bc --- /dev/null +++ b/Qwen_Qwen3-8B-Q4_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9bd6e5597ad3a70ef5b9a0acb995f739e545c5f16cbdf52b1c78835172b893c +size 4787332640 diff --git a/Qwen_Qwen3-8B-Q4_1.gguf b/Qwen_Qwen3-8B-Q4_1.gguf new file mode 100644 index 0000000..9291abd --- /dev/null +++ b/Qwen_Qwen3-8B-Q4_1.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebe71a958467349036e2c27be057473c17bb5fbd08700614ccb3aed8fb77880b +size 5247755808 diff --git a/Qwen_Qwen3-8B-Q4_K_L.gguf b/Qwen_Qwen3-8B-Q4_K_L.gguf new file mode 100644 index 0000000..a1fcdca --- /dev/null +++ b/Qwen_Qwen3-8B-Q4_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e59f21097a8102a681b72ea466b7c2389be92605844e5302ec3a9f20841f4f7 +size 5489669664 diff --git a/Qwen_Qwen3-8B-Q4_K_M.gguf b/Qwen_Qwen3-8B-Q4_K_M.gguf new file mode 100644 index 0000000..3ed7b3f --- /dev/null +++ b/Qwen_Qwen3-8B-Q4_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54fffa050078e984116639c83dfb64b5aa6d4cd474e018b076777c632bbccccd +size 5027784224 diff --git a/Qwen_Qwen3-8B-Q4_K_S.gguf b/Qwen_Qwen3-8B-Q4_K_S.gguf new file mode 100644 index 0000000..9311f4d --- /dev/null +++ b/Qwen_Qwen3-8B-Q4_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3672483a71bb93e2a18c1c1b8ca39e0b7018ce877ca8aff3df6539184a042fb0 +size 4802012704 diff --git a/Qwen_Qwen3-8B-Q5_K_L.gguf b/Qwen_Qwen3-8B-Q5_K_L.gguf new file mode 100644 index 0000000..796758c --- /dev/null +++ b/Qwen_Qwen3-8B-Q5_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0dcf4d30b6c92cf66d87cf1abea4e4a8b82aba3fa327e9a05e3c28152a12afee +size 6235207200 diff --git a/Qwen_Qwen3-8B-Q5_K_M.gguf b/Qwen_Qwen3-8B-Q5_K_M.gguf new file mode 100644 index 0000000..e60b767 --- /dev/null +++ b/Qwen_Qwen3-8B-Q5_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa45032e10b515a10374426bee3a41d5340aa78b67f0bed503469479a9772d9b +size 5851112992 diff --git a/Qwen_Qwen3-8B-Q5_K_S.gguf b/Qwen_Qwen3-8B-Q5_K_S.gguf new file mode 100644 index 0000000..864cf54 --- /dev/null +++ b/Qwen_Qwen3-8B-Q5_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ce0cb0736182baac640aaced416deb59b0a6a775712de1ec97002788291d00d +size 5720761888 diff --git a/Qwen_Qwen3-8B-Q6_K.gguf b/Qwen_Qwen3-8B-Q6_K.gguf new file mode 100644 index 0000000..07f4ec2 --- /dev/null +++ b/Qwen_Qwen3-8B-Q6_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69a45d3c366bab1736b201fdb21eb9d58160999a0f330a79d2c096f50de8985d +size 6725899808 diff --git a/Qwen_Qwen3-8B-Q6_K_L.gguf b/Qwen_Qwen3-8B-Q6_K_L.gguf new file mode 100644 index 0000000..ed1bbfa --- /dev/null +++ b/Qwen_Qwen3-8B-Q6_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2fdbe5817a1f7afb0207e2f0a6f336670a62dd5bd5090dc7d890edfda533208 +size 7027340832 diff --git a/Qwen_Qwen3-8B-Q8_0.gguf b/Qwen_Qwen3-8B-Q8_0.gguf new file mode 100644 index 0000000..e198b45 --- /dev/null +++ b/Qwen_Qwen3-8B-Q8_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edbd7ae01df991a2d4061e61451ad3d6829d8ca19f7cc846f7177499ac280c33 +size 8709518880 diff --git a/Qwen_Qwen3-8B-bf16.gguf b/Qwen_Qwen3-8B-bf16.gguf new file mode 100644 index 0000000..b730d04 --- /dev/null +++ b/Qwen_Qwen3-8B-bf16.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3eaf8da180bdfef9b16b22c4a9460872e8dcaaf315ee7e29413c9f8baa5fe6b8 +size 16388044064 diff --git a/Qwen_Qwen3-8B.imatrix b/Qwen_Qwen3-8B.imatrix new file mode 100644 index 0000000..a90eb49 --- /dev/null +++ b/Qwen_Qwen3-8B.imatrix @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cd37e7ed514b9ddc1eab833ca1ec21c841efc30ba788310389b4464c75653c5 +size 5316782 diff --git a/README.md b/README.md new file mode 100644 index 0000000..dfeadce --- /dev/null +++ b/README.md @@ -0,0 +1,171 @@ +--- +quantized_by: bartowski +pipeline_tag: text-generation +license: apache-2.0 +base_model: Qwen/Qwen3-8B +base_model_relation: quantized +--- + +## Llamacpp imatrix Quantizations of Qwen3-8B by Qwen + +Using llama.cpp release b5200 for quantization. + +Original model: https://huggingface.co/Qwen/Qwen3-8B + +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 + +``` +<|im_start|>system +{system_prompt}<|im_end|> +<|im_start|>user +{prompt}<|im_end|> +<|im_start|>assistant +``` + +## Download a file (not the whole branch) from below: + +| Filename | Quant type | File Size | Split | Description | +| -------- | ---------- | --------- | ----- | ----------- | +| [Qwen3-8B-bf16.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-bf16.gguf) | bf16 | 16.39GB | false | Full BF16 weights. | +| [Qwen3-8B-Q8_0.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q8_0.gguf) | Q8_0 | 8.71GB | false | Extremely high quality, generally unneeded but max available quant. | +| [Qwen3-8B-Q6_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q6_K_L.gguf) | Q6_K_L | 7.03GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. | +| [Qwen3-8B-Q6_K.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q6_K.gguf) | Q6_K | 6.73GB | false | Very high quality, near perfect, *recommended*. | +| [Qwen3-8B-Q5_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q5_K_L.gguf) | Q5_K_L | 6.24GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. | +| [Qwen3-8B-Q5_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q5_K_M.gguf) | Q5_K_M | 5.85GB | false | High quality, *recommended*. | +| [Qwen3-8B-Q5_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q5_K_S.gguf) | Q5_K_S | 5.72GB | false | High quality, *recommended*. | +| [Qwen3-8B-Q4_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q4_K_L.gguf) | Q4_K_L | 5.49GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. | +| [Qwen3-8B-Q4_1.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q4_1.gguf) | Q4_1 | 5.25GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. | +| [Qwen3-8B-Q4_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q4_K_M.gguf) | Q4_K_M | 5.03GB | false | Good quality, default size for most use cases, *recommended*. | +| [Qwen3-8B-Q3_K_XL.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q3_K_XL.gguf) | Q3_K_XL | 4.98GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. | +| [Qwen3-8B-Q4_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q4_K_S.gguf) | Q4_K_S | 4.80GB | false | Slightly lower quality with more space savings, *recommended*. | +| [Qwen3-8B-Q4_0.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q4_0.gguf) | Q4_0 | 4.79GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. | +| [Qwen3-8B-IQ4_NL.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ4_NL.gguf) | IQ4_NL | 4.79GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. | +| [Qwen3-8B-IQ4_XS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ4_XS.gguf) | IQ4_XS | 4.56GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. | +| [Qwen3-8B-Q3_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q3_K_L.gguf) | Q3_K_L | 4.43GB | false | Lower quality but usable, good for low RAM availability. | +| [Qwen3-8B-Q3_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q3_K_M.gguf) | Q3_K_M | 4.12GB | false | Low quality. | +| [Qwen3-8B-IQ3_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ3_M.gguf) | IQ3_M | 3.90GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. | +| [Qwen3-8B-Q2_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q2_K_L.gguf) | Q2_K_L | 3.89GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. | +| [Qwen3-8B-Q3_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q3_K_S.gguf) | Q3_K_S | 3.77GB | false | Low quality, not recommended. | +| [Qwen3-8B-IQ3_XS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ3_XS.gguf) | IQ3_XS | 3.63GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. | +| [Qwen3-8B-IQ3_XXS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ3_XXS.gguf) | IQ3_XXS | 3.37GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. | +| [Qwen3-8B-Q2_K.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-Q2_K.gguf) | Q2_K | 3.28GB | false | Very low quality but surprisingly usable. | +| [Qwen3-8B-IQ2_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-8B-GGUF/blob/main/Qwen_Qwen3-8B-IQ2_M.gguf) | IQ2_M | 3.05GB | 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/Qwen_Qwen3-8B-GGUF --include "Qwen_Qwen3-8B-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/Qwen_Qwen3-8B-GGUF --include "Qwen_Qwen3-8B-Q8_0/*" --local-dir ./ +``` + +You can either specify a new local-dir (Qwen_Qwen3-8B-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/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/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