commit 7853001a75425c3b221f452fe8e78d76b6dd0530 Author: ModelHub XC Date: Sat Jul 18 02:13:06 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: bartowski/Qwen_Qwen3-0.6B-GGUF Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..53d7257 --- /dev/null +++ b/.gitattributes @@ -0,0 +1,47 @@ +*.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 \ No newline at end of file diff --git a/Qwen_Qwen3-0.6B-IQ2_M.gguf b/Qwen_Qwen3-0.6B-IQ2_M.gguf new file mode 100644 index 0000000..47cbfe0 --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ2_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d129244d6f5285594084390f05ca5c3f7d194fd82f9d63f76c47e64725cf3b07 +size 331761056 diff --git a/Qwen_Qwen3-0.6B-IQ3_M.gguf b/Qwen_Qwen3-0.6B-IQ3_M.gguf new file mode 100644 index 0000000..729f2ab --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ3_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4efdacce21a52b1ca0018d7850f67562c59c27a1507d485cc1142bef1a09f44 +size 402878880 diff --git a/Qwen_Qwen3-0.6B-IQ3_XS.gguf b/Qwen_Qwen3-0.6B-IQ3_XS.gguf new file mode 100644 index 0000000..834ab73 --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ3_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0906dcf62aa4f005887e855731376f0bf3be8f5ffa61f547df71e6154ea748d9 +size 379605408 diff --git a/Qwen_Qwen3-0.6B-IQ3_XXS.gguf b/Qwen_Qwen3-0.6B-IQ3_XXS.gguf new file mode 100644 index 0000000..f876d28 --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ3_XXS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4553e20544d2667785db4e48b5556d02e70989234be5853b8bf200d65381d701 +size 345867680 diff --git a/Qwen_Qwen3-0.6B-IQ4_NL.gguf b/Qwen_Qwen3-0.6B-IQ4_NL.gguf new file mode 100644 index 0000000..147ef02 --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ4_NL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96c4d65dc596b294d7e74a5c849532836d009bc68ce763b80acb4995b4d30d09 +size 469081504 diff --git a/Qwen_Qwen3-0.6B-IQ4_XS.gguf b/Qwen_Qwen3-0.6B-IQ4_XS.gguf new file mode 100644 index 0000000..c5810b0 --- /dev/null +++ b/Qwen_Qwen3-0.6B-IQ4_XS.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87c112e51ae42471c51081e9ea35c7ec57d6b93d72d857affbc84790eda31341 +size 450456992 diff --git a/Qwen_Qwen3-0.6B-Q2_K.gguf b/Qwen_Qwen3-0.6B-Q2_K.gguf new file mode 100644 index 0000000..e5cfc14 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q2_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1960f01d2aa0d8c7d4e5cb13026e4c3dc85c899c0851bddb2cabbbde18ac52b +size 347288992 diff --git a/Qwen_Qwen3-0.6B-Q2_K_L.gguf b/Qwen_Qwen3-0.6B-Q2_K_L.gguf new file mode 100644 index 0000000..1c8ce93 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q2_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e2c3e043d29a52dad50003fde24f201cf4006b3e5dd7e45dcde270cd180d9d2 +size 499224992 diff --git a/Qwen_Qwen3-0.6B-Q3_K_L.gguf b/Qwen_Qwen3-0.6B-Q3_K_L.gguf new file mode 100644 index 0000000..7735adb --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q3_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0793e0df451c226d5229fb9728624c5359d539ffaab3f6a0d28935eb09f8e016 +size 435343776 diff --git a/Qwen_Qwen3-0.6B-Q3_K_M.gguf b/Qwen_Qwen3-0.6B-Q3_K_M.gguf new file mode 100644 index 0000000..d965d96 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q3_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81771904092305dcf02980cb7705111c7ec0c8b8cffd497a7d05964d3373a45e +size 413979040 diff --git a/Qwen_Qwen3-0.6B-Q3_K_S.gguf b/Qwen_Qwen3-0.6B-Q3_K_S.gguf new file mode 100644 index 0000000..d1fc650 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q3_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6552632340ff82d8ea6f519c384eede4dd37933b200a05890a65dabae612327 +size 389927328 diff --git a/Qwen_Qwen3-0.6B-Q3_K_XL.gguf b/Qwen_Qwen3-0.6B-Q3_K_XL.gguf new file mode 100644 index 0000000..32f8f8b --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q3_K_XL.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03f95d27a9d8374b1298d80bef80a5b33f93cd943e7806139239d99d6e2be092 +size 571478432 diff --git a/Qwen_Qwen3-0.6B-Q4_0.gguf b/Qwen_Qwen3-0.6B-Q4_0.gguf new file mode 100644 index 0000000..7f935f7 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q4_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b78d8e3c61976cebb78ef5affe19d0eca75b1b47ec66f613a5a3245484758d5 +size 469671328 diff --git a/Qwen_Qwen3-0.6B-Q4_1.gguf b/Qwen_Qwen3-0.6B-Q4_1.gguf new file mode 100644 index 0000000..af5162d --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q4_1.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:459dad8b75080ce0c96ba2ed7514887f942bd9739c40aad0923ba324c34fb25e +size 506330528 diff --git a/Qwen_Qwen3-0.6B-Q4_K_L.gguf b/Qwen_Qwen3-0.6B-Q4_K_L.gguf new file mode 100644 index 0000000..b1994a9 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q4_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:109aa91357a258782343e60edf46b83e1403652af1b91635b4f1c91b4a69ddc6 +size 599691680 diff --git a/Qwen_Qwen3-0.6B-Q4_K_M.gguf b/Qwen_Qwen3-0.6B-Q4_K_M.gguf new file mode 100644 index 0000000..4c42fc7 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q4_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9acfc1e001311f34b4252001b626f2e466d592a42065f66571bff3790d4e1b14 +size 484220320 diff --git a/Qwen_Qwen3-0.6B-Q4_K_S.gguf b/Qwen_Qwen3-0.6B-Q4_K_S.gguf new file mode 100644 index 0000000..684e306 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q4_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64d751138caefeb5a6c8d8839598ce66916b16a47b9be54b12b2ad78a3f6b4ef +size 470785440 diff --git a/Qwen_Qwen3-0.6B-Q5_K_L.gguf b/Qwen_Qwen3-0.6B-Q5_K_L.gguf new file mode 100644 index 0000000..504b12e --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q5_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74e71eae7916b4c007e5c48f8ae477292f067c69060a4202dfc25d85b23baf91 +size 647401888 diff --git a/Qwen_Qwen3-0.6B-Q5_K_M.gguf b/Qwen_Qwen3-0.6B-Q5_K_M.gguf new file mode 100644 index 0000000..04cf699 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q5_K_M.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2df7894c99ee7716f0ed3c7d454c2386a7ae976c4b0ad0e716d55399c1360648 +size 551378336 diff --git a/Qwen_Qwen3-0.6B-Q5_K_S.gguf b/Qwen_Qwen3-0.6B-Q5_K_S.gguf new file mode 100644 index 0000000..61585bb --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q5_K_S.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43c5efb14850c6aa2f4c70384fe6e8079d9cf5a9f457e68f9a25c9a1aaa6a04b +size 543579552 diff --git a/Qwen_Qwen3-0.6B-Q6_K.gguf b/Qwen_Qwen3-0.6B-Q6_K.gguf new file mode 100644 index 0000000..7ed0ba9 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q6_K.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21e95febea68b223089e2382378e8c3155e587f5a6618ae3595e3d00c6885575 +size 622733728 diff --git a/Qwen_Qwen3-0.6B-Q6_K_L.gguf b/Qwen_Qwen3-0.6B-Q6_K_L.gguf new file mode 100644 index 0000000..feaeb2f --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q6_K_L.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3356ceb1dc0916ceb45022b50351cc7341aeaa3b15ec89e71fc4194f603b538 +size 698093984 diff --git a/Qwen_Qwen3-0.6B-Q8_0.gguf b/Qwen_Qwen3-0.6B-Q8_0.gguf new file mode 100644 index 0000000..f664e22 --- /dev/null +++ b/Qwen_Qwen3-0.6B-Q8_0.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c159d1518f16bc42533d9a09f034eb598b670341adc2b08b9a9751614aea71eb +size 804753824 diff --git a/Qwen_Qwen3-0.6B-bf16.gguf b/Qwen_Qwen3-0.6B-bf16.gguf new file mode 100644 index 0000000..f18fd10 --- /dev/null +++ b/Qwen_Qwen3-0.6B-bf16.gguf @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6bbe7c978ab2dab9f4e70c040ef4ad53928cacf3d24bc5dc7252aff93be92386 +size 1509347488 diff --git a/Qwen_Qwen3-0.6B.imatrix b/Qwen_Qwen3-0.6B.imatrix new file mode 100644 index 0000000..8000834 Binary files /dev/null and b/Qwen_Qwen3-0.6B.imatrix differ diff --git a/README.md b/README.md new file mode 100644 index 0000000..22b9d8d --- /dev/null +++ b/README.md @@ -0,0 +1,170 @@ +--- +quantized_by: bartowski +pipeline_tag: text-generation +base_model: Qwen/Qwen3-0.6B +base_model_relation: quantized +--- + +## Llamacpp imatrix Quantizations of Qwen3-0.6B by Qwen + +Using llama.cpp release b5200 for quantization. + +Original model: https://huggingface.co/Qwen/Qwen3-0.6B + +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 | +| -------- | ---------- | --------- | ----- | ----------- | +| [Qwen3-0.6B-bf16.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-bf16.gguf) | bf16 | 1.51GB | false | Full BF16 weights. | +| [Qwen3-0.6B-Q8_0.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q8_0.gguf) | Q8_0 | 0.80GB | false | Extremely high quality, generally unneeded but max available quant. | +| [Qwen3-0.6B-Q6_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q6_K_L.gguf) | Q6_K_L | 0.70GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. | +| [Qwen3-0.6B-Q5_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q5_K_L.gguf) | Q5_K_L | 0.65GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. | +| [Qwen3-0.6B-Q6_K.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q6_K.gguf) | Q6_K | 0.62GB | false | Very high quality, near perfect, *recommended*. | +| [Qwen3-0.6B-Q4_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q4_K_L.gguf) | Q4_K_L | 0.60GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. | +| [Qwen3-0.6B-Q3_K_XL.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q3_K_XL.gguf) | Q3_K_XL | 0.57GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. | +| [Qwen3-0.6B-Q5_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q5_K_M.gguf) | Q5_K_M | 0.55GB | false | High quality, *recommended*. | +| [Qwen3-0.6B-Q5_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q5_K_S.gguf) | Q5_K_S | 0.54GB | false | High quality, *recommended*. | +| [Qwen3-0.6B-Q4_1.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q4_1.gguf) | Q4_1 | 0.51GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. | +| [Qwen3-0.6B-Q2_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q2_K_L.gguf) | Q2_K_L | 0.50GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. | +| [Qwen3-0.6B-Q4_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q4_K_M.gguf) | Q4_K_M | 0.48GB | false | Good quality, default size for most use cases, *recommended*. | +| [Qwen3-0.6B-Q4_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q4_K_S.gguf) | Q4_K_S | 0.47GB | false | Slightly lower quality with more space savings, *recommended*. | +| [Qwen3-0.6B-Q4_0.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q4_0.gguf) | Q4_0 | 0.47GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. | +| [Qwen3-0.6B-IQ4_NL.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ4_NL.gguf) | IQ4_NL | 0.47GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. | +| [Qwen3-0.6B-IQ4_XS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ4_XS.gguf) | IQ4_XS | 0.45GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. | +| [Qwen3-0.6B-Q3_K_L.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q3_K_L.gguf) | Q3_K_L | 0.44GB | false | Lower quality but usable, good for low RAM availability. | +| [Qwen3-0.6B-Q3_K_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q3_K_M.gguf) | Q3_K_M | 0.41GB | false | Low quality. | +| [Qwen3-0.6B-IQ3_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ3_M.gguf) | IQ3_M | 0.40GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. | +| [Qwen3-0.6B-Q3_K_S.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q3_K_S.gguf) | Q3_K_S | 0.39GB | false | Low quality, not recommended. | +| [Qwen3-0.6B-IQ3_XS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ3_XS.gguf) | IQ3_XS | 0.38GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. | +| [Qwen3-0.6B-IQ3_XXS.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ3_XXS.gguf) | IQ3_XXS | 0.35GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. | +| [Qwen3-0.6B-Q2_K.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-Q2_K.gguf) | Q2_K | 0.35GB | false | Very low quality but surprisingly usable. | +| [Qwen3-0.6B-IQ2_M.gguf](https://huggingface.co/bartowski/Qwen_Qwen3-0.6B-GGUF/blob/main/Qwen_Qwen3-0.6B-IQ2_M.gguf) | IQ2_M | 0.33GB | 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_Qwen3-0.6B-GGUF --include "Qwen_Qwen3-0.6B-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_Qwen3-0.6B-GGUF --include "Qwen_Qwen3-0.6B-Q8_0/*" --local-dir ./ +``` + +You can either specify a new local-dir (Qwen_Qwen3-0.6B-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