Hugging Face's logo

XelotX
/
QVQ-72B-Preview-GGUF

Image-Text-to-Text
GGUF
English
chat
Inference Endpoints
imatrix
conversational
Model card Files Community
XelotX bartowski commited on
Commit
8004c0c
·
verified ·
0 Parent(s):

Duplicate from bartowski/QVQ-72B-Preview-GGUF

Browse files

Co-authored-by: Bartowski <[email protected]>

Files changed (35) hide show
  1. .gitattributes +68 -0
  2. QVQ-72B-Preview-IQ1_M.gguf +3 -0
  3. QVQ-72B-Preview-IQ2_M.gguf +3 -0
  4. QVQ-72B-Preview-IQ2_S.gguf +3 -0
  5. QVQ-72B-Preview-IQ2_XS.gguf +3 -0
  6. QVQ-72B-Preview-IQ2_XXS.gguf +3 -0
  7. QVQ-72B-Preview-IQ3_M.gguf +3 -0
  8. QVQ-72B-Preview-IQ3_XXS.gguf +3 -0
  9. QVQ-72B-Preview-IQ4_NL.gguf +3 -0
  10. QVQ-72B-Preview-IQ4_XS.gguf +3 -0
  11. QVQ-72B-Preview-Q2_K.gguf +3 -0
  12. QVQ-72B-Preview-Q2_K_L.gguf +3 -0
  13. QVQ-72B-Preview-Q3_K_L.gguf +3 -0
  14. QVQ-72B-Preview-Q3_K_M.gguf +3 -0
  15. QVQ-72B-Preview-Q3_K_S.gguf +3 -0
  16. QVQ-72B-Preview-Q3_K_XL.gguf +3 -0
  17. QVQ-72B-Preview-Q4_0.gguf +3 -0
  18. QVQ-72B-Preview-Q4_1.gguf +3 -0
  19. QVQ-72B-Preview-Q4_K_L.gguf +3 -0
  20. QVQ-72B-Preview-Q4_K_M.gguf +3 -0
  21. QVQ-72B-Preview-Q4_K_S.gguf +3 -0
  22. QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00001-of-00002.gguf +3 -0
  23. QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00002-of-00002.gguf +3 -0
  24. QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00001-of-00002.gguf +3 -0
  25. QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00002-of-00002.gguf +3 -0
  26. QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00001-of-00002.gguf +3 -0
  27. QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00002-of-00002.gguf +3 -0
  28. QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00001-of-00002.gguf +3 -0
  29. QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00002-of-00002.gguf +3 -0
  30. QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00001-of-00002.gguf +3 -0
  31. QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00002-of-00002.gguf +3 -0
  32. QVQ-72B-Preview.imatrix +3 -0
  33. README.md +184 -0
  34. mmproj-QVQ-72B-Preview-f16.gguf +3 -0
  35. mmproj-QVQ-72B-Preview-f32.gguf +3 -0
.gitattributes ADDED
@@ -0,0 +1,68 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
37
+ QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
38
+ QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
39
+ QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
40
+ QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
41
+ QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
42
+ QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
43
+ QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
44
+ QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
45
+ QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
46
+ QVQ-72B-Preview-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text
47
+ QVQ-72B-Preview-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
48
+ QVQ-72B-Preview-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
49
+ QVQ-72B-Preview-Q4_1.gguf filter=lfs diff=lfs merge=lfs -text
50
+ QVQ-72B-Preview-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
51
+ QVQ-72B-Preview-IQ4_NL.gguf filter=lfs diff=lfs merge=lfs -text
52
+ QVQ-72B-Preview-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text
53
+ QVQ-72B-Preview-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
54
+ QVQ-72B-Preview-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text
55
+ QVQ-72B-Preview-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
56
+ QVQ-72B-Preview-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text
57
+ QVQ-72B-Preview-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
58
+ QVQ-72B-Preview-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text
59
+ QVQ-72B-Preview-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text
60
+ QVQ-72B-Preview-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text
61
+ QVQ-72B-Preview-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text
62
+ QVQ-72B-Preview-IQ2_S.gguf filter=lfs diff=lfs merge=lfs -text
63
+ QVQ-72B-Preview-IQ2_XS.gguf filter=lfs diff=lfs merge=lfs -text
64
+ QVQ-72B-Preview-IQ2_XXS.gguf filter=lfs diff=lfs merge=lfs -text
65
+ QVQ-72B-Preview-IQ1_M.gguf filter=lfs diff=lfs merge=lfs -text
66
+ QVQ-72B-Preview.imatrix filter=lfs diff=lfs merge=lfs -text
67
+ mmproj-QVQ-72B-Preview-f16.gguf filter=lfs diff=lfs merge=lfs -text
68
+ mmproj-QVQ-72B-Preview-f32.gguf filter=lfs diff=lfs merge=lfs -text
QVQ-72B-Preview-IQ1_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bdd86103bf379746abbd5423cbe9f45b9aee10194ca6910d44fa403897db8874
3
+ size 23740211552
QVQ-72B-Preview-IQ2_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86b303bdbca169a24f993e356218e0e2c4f75c5530db66ad3deda65e195f1b28
3
+ size 29338984800
QVQ-72B-Preview-IQ2_S.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eba21c09f79f07e1a2d5a32b516088f9f33228a1e4767110fb41b3f7c9564604
3
+ size 27939135840
QVQ-72B-Preview-IQ2_XS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d367c2ba5dedaa5506198157f16d91161129068331024b6a00b154cba89d7dbb
3
+ size 27057643872
QVQ-72B-Preview-IQ2_XXS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca7deb07da9198cf5100127adc421a0e34db1cf641cda9f12379f7b769a9c5ae
3
+ size 25490022752
QVQ-72B-Preview-IQ3_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e612b5235b439ff5f0e196dab71811335d1118a29517561edefaf40958d0fa09
3
+ size 35503595872
QVQ-72B-Preview-IQ3_XXS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8df03242760ca188bd2f927a4d79dab1e1189c964385b78e62b974d438f02781
3
+ size 31845081440
QVQ-72B-Preview-IQ4_NL.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:571e77a1c5affb473a42fa2316c4392b2e81fdb2e8f6fd8bc67bba5eb3fb8b30
3
+ size 41315623264
QVQ-72B-Preview-IQ4_XS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0768fdf6c923f9d5af2de58bc9c89dba20c35079a35b7d3bce618b15fa1f3f01
3
+ size 39709073760
QVQ-72B-Preview-Q2_K.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5dcb6606a5ac69a76d2a02203e1bbd7c2a0c3a5dd09ec69dfa7af7f432bd1e43
3
+ size 29811761504
QVQ-72B-Preview-Q2_K_L.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b46ad56d36f8cb632dc867e1079a764e7f6a1b4abf72a6034be95ef382912b4b
3
+ size 31028273504
QVQ-72B-Preview-Q3_K_L.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56b328bb94dbd72dd1b817946507f0ef6fb843df6da7284601d8ad583a76445c
3
+ size 39505224032
QVQ-72B-Preview-Q3_K_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82820a36f030e7045e786284732edf0169544bd29be73c26736321af92bafdc4
3
+ size 37698724192
QVQ-72B-Preview-Q3_K_S.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:faedc9a8e3c5644da1a0bbbf3b9dbdeaaf88301c3b9e695eb0b5e3892f0a88ce
3
+ size 34487787872
QVQ-72B-Preview-Q3_K_XL.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c2cce9bb62cdfdd9cd23761f5e781c6fb1215b223ae98ca6d5dca80c83810af
3
+ size 40595218784
QVQ-72B-Preview-Q4_0.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:371fdb3f0d9d268012a06396e4b05c0791ec81e6e0fd83c857f4b4e2aca958ed
3
+ size 41383125344
QVQ-72B-Preview-Q4_1.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97542bcdff41d1e0f63fb682e3fa4fe8ae958382417862daab0daedc974ebfba
3
+ size 45697884512
QVQ-72B-Preview-Q4_K_L.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49ecab6afc82486151830989cb79d3bf58b88197c714197968c7fd2f4523afe3
3
+ size 48340263264
QVQ-72B-Preview-Q4_K_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0fab6809995614c19e4b4c23e3191824944a04999f742486278f0d9929dc82ae
3
+ size 47415714144
QVQ-72B-Preview-Q4_K_S.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a632c50ab1ad776111338791c4a877704d50cf385e2c9e3512685da6b4c0c46
3
+ size 43889221984
QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45d786e3db4fa5f42ab22ef48d376659563982b9e6967ec4c1e52172cdf1c1fe
3
+ size 39940548000
QVQ-72B-Preview-Q5_K_L/QVQ-72B-Preview-Q5_K_L-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0893bd03a583b383688bb7b968ac04ed9c0a2da573f7a534ec804cb5e8600727
3
+ size 15275752576
QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:699cfd5dae7c608f7022a6e28020c30a927c5766db6839c825f8e2d2f9ffedfa
3
+ size 39866042464
QVQ-72B-Preview-Q5_K_M/QVQ-72B-Preview-Q5_K_M-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:16d0de991c4262303ac628e4f01f39b60a7aed904511db3aa24cb002d0ccf7fa
3
+ size 14581422560
QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1330d2c021af85c223a66a17becc5c8f31336685fad22f8e317ee9484dd960f
3
+ size 39906603168
QVQ-72B-Preview-Q5_K_S/QVQ-72B-Preview-Q5_K_S-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ea27e201019dc690cac5c2c087673b4fa9b1ce5fe2a1d270dfbce6fa892d1d0
3
+ size 11468534144
QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57def665762cf0495fcfc9b7e74efe89cca7f6d053ec1c32b4ad0933506b7f8f
3
+ size 39809756608
QVQ-72B-Preview-Q6_K/QVQ-72B-Preview-Q6_K-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68da46bb82817462bf7258258b6e66ff4d87daa444a899ce39aac1804c3117a4
3
+ size 24537871488
QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0181fb75247f6836dcd2d5f0c50bc2f5d69743c242eae2ee9d558ba0ec346a98
3
+ size 39966803392
QVQ-72B-Preview-Q8_0/QVQ-72B-Preview-Q8_0-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1acf6af0776b2475f05990897696b281e2c66ac552e232ddae9fd8818922ef9e
3
+ size 37295807328
QVQ-72B-Preview.imatrix ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f506300b7740760944d242bf9ba89bbbda3b3aecfe14eb10e7aa73b9e9283d6e
3
+ size 25209018
README.md ADDED
@@ -0,0 +1,184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ quantized_by: bartowski
3
+ pipeline_tag: image-text-to-text
4
+ license_name: qwen
5
+ language:
6
+ - en
7
+ license_link: https://huggingface.co/Qwen/QVQ-72B-Preview/blob/main/LICENSE
8
+ base_model: Qwen/QVQ-72B-Preview
9
+ license: other
10
+ tags:
11
+ - chat
12
+ ---
13
+
14
+ ## Llamacpp imatrix Quantizations of QVQ-72B-Preview
15
+
16
+ Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b4381">b4381</a> for quantization.
17
+
18
+ Original model: https://huggingface.co/Qwen/QVQ-72B-Preview
19
+
20
+ All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8)
21
+
22
+ ## How to run
23
+
24
+ QwenVL models need to use the `llama-qwen2vl-cli` tool like so:
25
+
26
+ ```
27
+ ./llama-qwen2vl-cli -m /models/QVQ-72B-Preview-Q4_K_M.gguf \
28
+ --mmproj /models/mmproj-QVQ-72B-Preview-f16.gguf \
29
+ -p 'How many fingers does this hand have.' \
30
+ --image '/models/hand.jpg'
31
+ ```
32
+
33
+ ## Prompt format
34
+
35
+ ```
36
+ <|im_start|>system
37
+ {system_prompt}<|im_end|>
38
+ <|im_start|>user
39
+ {prompt}<|im_end|>
40
+ <|im_start|>assistant
41
+ ```
42
+
43
+ ## Download a file (not the whole branch) from below:
44
+
45
+ | Filename | Quant type | File Size | Split | Description |
46
+ | -------- | ---------- | --------- | ----- | ----------- |
47
+ | [QVQ-72B-Preview-Q8_0.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/tree/main/QVQ-72B-Preview-Q8_0) | Q8_0 | 77.26GB | true | Extremely high quality, generally unneeded but max available quant. |
48
+ | [QVQ-72B-Preview-Q6_K.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/tree/main/QVQ-72B-Preview-Q6_K) | Q6_K | 64.35GB | true | Very high quality, near perfect, *recommended*. |
49
+ | [QVQ-72B-Preview-Q5_K_L.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/tree/main/QVQ-72B-Preview-Q5_K_L) | Q5_K_L | 55.22GB | true | Uses Q8_0 for embed and output weights. High quality, *recommended*. |
50
+ | [QVQ-72B-Preview-Q5_K_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/tree/main/QVQ-72B-Preview-Q5_K_M) | Q5_K_M | 54.45GB | true | High quality, *recommended*. |
51
+ | [QVQ-72B-Preview-Q5_K_S.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/tree/main/QVQ-72B-Preview-Q5_K_S) | Q5_K_S | 51.38GB | true | High quality, *recommended*. |
52
+ | [QVQ-72B-Preview-Q4_K_L.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q4_K_L.gguf) | Q4_K_L | 48.34GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. |
53
+ | [QVQ-72B-Preview-Q4_K_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q4_K_M.gguf) | Q4_K_M | 47.42GB | false | Good quality, default size for most use cases, *recommended*. |
54
+ | [QVQ-72B-Preview-Q4_1.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q4_1.gguf) | Q4_1 | 45.70GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. |
55
+ | [QVQ-72B-Preview-Q4_K_S.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q4_K_S.gguf) | Q4_K_S | 43.89GB | false | Slightly lower quality with more space savings, *recommended*. |
56
+ | [QVQ-72B-Preview-Q4_0.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q4_0.gguf) | Q4_0 | 41.38GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. |
57
+ | [QVQ-72B-Preview-IQ4_NL.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ4_NL.gguf) | IQ4_NL | 41.32GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. |
58
+ | [QVQ-72B-Preview-Q3_K_XL.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q3_K_XL.gguf) | Q3_K_XL | 40.60GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
59
+ | [QVQ-72B-Preview-IQ4_XS.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ4_XS.gguf) | IQ4_XS | 39.71GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
60
+ | [QVQ-72B-Preview-Q3_K_L.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q3_K_L.gguf) | Q3_K_L | 39.51GB | false | Lower quality but usable, good for low RAM availability. |
61
+ | [QVQ-72B-Preview-Q3_K_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q3_K_M.gguf) | Q3_K_M | 37.70GB | false | Low quality. |
62
+ | [QVQ-72B-Preview-IQ3_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ3_M.gguf) | IQ3_M | 35.50GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
63
+ | [QVQ-72B-Preview-Q3_K_S.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q3_K_S.gguf) | Q3_K_S | 34.49GB | false | Low quality, not recommended. |
64
+ | [QVQ-72B-Preview-IQ3_XXS.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ3_XXS.gguf) | IQ3_XXS | 31.85GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. |
65
+ | [QVQ-72B-Preview-Q2_K_L.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q2_K_L.gguf) | Q2_K_L | 31.03GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
66
+ | [QVQ-72B-Preview-Q2_K.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-Q2_K.gguf) | Q2_K | 29.81GB | false | Very low quality but surprisingly usable. |
67
+ | [QVQ-72B-Preview-IQ2_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ2_M.gguf) | IQ2_M | 29.34GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
68
+ | [QVQ-72B-Preview-IQ2_S.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ2_S.gguf) | IQ2_S | 27.94GB | false | Low quality, uses SOTA techniques to be usable. |
69
+ | [QVQ-72B-Preview-IQ2_XS.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ2_XS.gguf) | IQ2_XS | 27.06GB | false | Low quality, uses SOTA techniques to be usable. |
70
+ | [QVQ-72B-Preview-IQ2_XXS.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ2_XXS.gguf) | IQ2_XXS | 25.49GB | false | Very low quality, uses SOTA techniques to be usable. |
71
+ | [QVQ-72B-Preview-IQ1_M.gguf](https://huggingface.co/bartowski/QVQ-72B-Preview-GGUF/blob/main/QVQ-72B-Preview-IQ1_M.gguf) | IQ1_M | 23.74GB | false | Extremely low quality, *not* recommended. |
72
+
73
+ ## Embed/output weights
74
+
75
+ 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.
76
+
77
+ ## Downloading using huggingface-cli
78
+
79
+ <details>
80
+ <summary>Click to view download instructions</summary>
81
+
82
+ First, make sure you have hugginface-cli installed:
83
+
84
+ ```
85
+ pip install -U "huggingface_hub[cli]"
86
+ ```
87
+
88
+ Then, you can target the specific file you want:
89
+
90
+ ```
91
+ huggingface-cli download bartowski/QVQ-72B-Preview-GGUF --include "QVQ-72B-Preview-Q4_K_M.gguf" --local-dir ./
92
+ ```
93
+
94
+ 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:
95
+
96
+ ```
97
+ huggingface-cli download bartowski/QVQ-72B-Preview-GGUF --include "QVQ-72B-Preview-Q8_0/*" --local-dir ./
98
+ ```
99
+
100
+ You can either specify a new local-dir (QVQ-72B-Preview-Q8_0) or download them all in place (./)
101
+
102
+ </details>
103
+
104
+ ## ARM/AVX information
105
+
106
+ 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.
107
+
108
+ 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.
109
+
110
+ 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.
111
+
112
+ 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.
113
+
114
+ <details>
115
+ <summary>Click to view Q4_0_X_X information (deprecated</summary>
116
+
117
+ I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking.
118
+
119
+ <details>
120
+ <summary>Click to view benchmarks on an AVX2 system (EPYC7702)</summary>
121
+
122
+ | model | size | params | backend | threads | test | t/s | % (vs Q4_0) |
123
+ | ------------------------------ | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |-------------: |
124
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp512 | 204.03 ± 1.03 | 100% |
125
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp1024 | 282.92 ± 0.19 | 100% |
126
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp2048 | 259.49 ± 0.44 | 100% |
127
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg128 | 39.12 ± 0.27 | 100% |
128
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg256 | 39.31 ± 0.69 | 100% |
129
+ | qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg512 | 40.52 ± 0.03 | 100% |
130
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp512 | 301.02 ± 1.74 | 147% |
131
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp1024 | 287.23 ± 0.20 | 101% |
132
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp2048 | 262.77 ± 1.81 | 101% |
133
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg128 | 18.80 ± 0.99 | 48% |
134
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg256 | 24.46 ± 3.04 | 83% |
135
+ | qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg512 | 36.32 ± 3.59 | 90% |
136
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp512 | 271.71 ± 3.53 | 133% |
137
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp1024 | 279.86 ± 45.63 | 100% |
138
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp2048 | 320.77 ± 5.00 | 124% |
139
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg128 | 43.51 ± 0.05 | 111% |
140
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg256 | 43.35 ± 0.09 | 110% |
141
+ | qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg512 | 42.60 ± 0.31 | 105% |
142
+
143
+ Q4_0_8_8 offers a nice bump to prompt processing and a small bump to text generation
144
+
145
+ </details>
146
+
147
+ </details>
148
+
149
+ ## Which file should I choose?
150
+
151
+ <details>
152
+ <summary>Click here for details</summary>
153
+
154
+ A great write up with charts showing various performances is provided by Artefact2 [here](https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9)
155
+
156
+ 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.
157
+
158
+ 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.
159
+
160
+ 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.
161
+
162
+ Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.
163
+
164
+ 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.
165
+
166
+ If you want to get more into the weeds, you can check out this extremely useful feature chart:
167
+
168
+ [llama.cpp feature matrix](https://github.com/ggerganov/llama.cpp/wiki/Feature-matrix)
169
+
170
+ 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.
171
+
172
+ These I-quants can also be used on CPU and Apple Metal, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.
173
+
174
+ The I-quants are *not* compatible with Vulcan, which is also AMD, so if you have an AMD card double check if you're using the rocBLAS build or the Vulcan build. At the time of writing this, LM Studio has a preview with ROCm support, and other inference engines have specific builds for ROCm.
175
+
176
+ </details>
177
+
178
+ ## Credits
179
+
180
+ Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.
181
+
182
+ Thank you ZeroWw for the inspiration to experiment with embed/output.
183
+
184
+ Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
mmproj-QVQ-72B-Preview-f16.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85110223f39aa1aad887052d269074afbd52a49ae02c53b66753b033662cc8e6
3
+ size 1399840256
mmproj-QVQ-72B-Preview-f32.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44ef508dde5c8578fd66789ccb99a25ea6caf2cab06b9b450a4859260007f2b6
3
+ size 2797456896