Hugging Face's logo

AXERA-TECH
/
Qwen3-VL-4B-Instruct

Image-Text-to-Text
Transformers
English
Chinese
Qwen3-VL
Qwen3-VL-2B-Instruct
Qwen3-VL-4B-Instruct
Int8
VLM
Model card Files
xet
Community
lihongjie commited on
Commit
7394e58
·
1 Parent(s): 3551b10

first commit

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +58 -0
  2. Qwen3-VL-4B-Instruct-AX650-c128_p1152/Qwen3-VL-4B-Instruct_vision.axmodel +3 -0
  3. Qwen3-VL-4B-Instruct-AX650-c128_p1152/model.embed_tokens.weight.bfloat16.bin +3 -0
  4. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l0_together.axmodel +3 -0
  5. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l10_together.axmodel +3 -0
  6. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l11_together.axmodel +3 -0
  7. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l12_together.axmodel +3 -0
  8. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l13_together.axmodel +3 -0
  9. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l14_together.axmodel +3 -0
  10. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l15_together.axmodel +3 -0
  11. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l16_together.axmodel +3 -0
  12. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l17_together.axmodel +3 -0
  13. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l18_together.axmodel +3 -0
  14. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l19_together.axmodel +3 -0
  15. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l1_together.axmodel +3 -0
  16. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l20_together.axmodel +3 -0
  17. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l21_together.axmodel +3 -0
  18. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l22_together.axmodel +3 -0
  19. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l23_together.axmodel +3 -0
  20. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l24_together.axmodel +3 -0
  21. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l25_together.axmodel +3 -0
  22. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l26_together.axmodel +3 -0
  23. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l27_together.axmodel +3 -0
  24. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l28_together.axmodel +3 -0
  25. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l29_together.axmodel +3 -0
  26. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l2_together.axmodel +3 -0
  27. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l30_together.axmodel +3 -0
  28. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l31_together.axmodel +3 -0
  29. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l32_together.axmodel +3 -0
  30. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l33_together.axmodel +3 -0
  31. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l34_together.axmodel +3 -0
  32. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l35_together.axmodel +3 -0
  33. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l3_together.axmodel +3 -0
  34. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l4_together.axmodel +3 -0
  35. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l5_together.axmodel +3 -0
  36. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l6_together.axmodel +3 -0
  37. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l7_together.axmodel +3 -0
  38. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l8_together.axmodel +3 -0
  39. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l9_together.axmodel +3 -0
  40. Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_post.axmodel +3 -0
  41. README.md +233 -3
  42. config.json +0 -0
  43. images/demo.jpg +3 -0
  44. images/demo1.jpg +3 -0
  45. images/recoAll_attractions_1.jpg +3 -0
  46. images/recoAll_attractions_2.jpg +3 -0
  47. images/recoAll_attractions_3.jpg +3 -0
  48. images/recoAll_attractions_4.jpg +3 -0
  49. images/ssd_car.jpg +3 -0
  50. images/ssd_horse.jpg +3 -0
.gitattributes CHANGED
@@ -33,3 +33,61 @@ saved_model/**/* 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
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ images/ssd_horse.jpg filter=lfs diff=lfs merge=lfs -text
37
+ video/frame_0000.jpg filter=lfs diff=lfs merge=lfs -text
38
+ video/frame_0032.jpg filter=lfs diff=lfs merge=lfs -text
39
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/model.embed_tokens.weight.bfloat16.bin filter=lfs diff=lfs merge=lfs -text
40
+ images/demo1.jpg filter=lfs diff=lfs merge=lfs -text
41
+ images/recoAll_attractions_3.jpg filter=lfs diff=lfs merge=lfs -text
42
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l20_together.axmodel filter=lfs diff=lfs merge=lfs -text
43
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l30_together.axmodel filter=lfs diff=lfs merge=lfs -text
44
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l8_together.axmodel filter=lfs diff=lfs merge=lfs -text
45
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l35_together.axmodel filter=lfs diff=lfs merge=lfs -text
46
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l6_together.axmodel filter=lfs diff=lfs merge=lfs -text
47
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_post.axmodel filter=lfs diff=lfs merge=lfs -text
48
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l12_together.axmodel filter=lfs diff=lfs merge=lfs -text
49
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l13_together.axmodel filter=lfs diff=lfs merge=lfs -text
50
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l17_together.axmodel filter=lfs diff=lfs merge=lfs -text
51
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l22_together.axmodel filter=lfs diff=lfs merge=lfs -text
52
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l34_together.axmodel filter=lfs diff=lfs merge=lfs -text
53
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l3_together.axmodel filter=lfs diff=lfs merge=lfs -text
54
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l15_together.axmodel filter=lfs diff=lfs merge=lfs -text
55
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l27_together.axmodel filter=lfs diff=lfs merge=lfs -text
56
+ video/frame_0008.jpg filter=lfs diff=lfs merge=lfs -text
57
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/Qwen3-VL-4B-Instruct_vision.axmodel filter=lfs diff=lfs merge=lfs -text
58
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l5_together.axmodel filter=lfs diff=lfs merge=lfs -text
59
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l2_together.axmodel filter=lfs diff=lfs merge=lfs -text
60
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l33_together.axmodel filter=lfs diff=lfs merge=lfs -text
61
+ images/demo.jpg filter=lfs diff=lfs merge=lfs -text
62
+ images/ssd_car.jpg filter=lfs diff=lfs merge=lfs -text
63
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l16_together.axmodel filter=lfs diff=lfs merge=lfs -text
64
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l18_together.axmodel filter=lfs diff=lfs merge=lfs -text
65
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l24_together.axmodel filter=lfs diff=lfs merge=lfs -text
66
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l28_together.axmodel filter=lfs diff=lfs merge=lfs -text
67
+ main_axcl_x86 filter=lfs diff=lfs merge=lfs -text
68
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l10_together.axmodel filter=lfs diff=lfs merge=lfs -text
69
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l11_together.axmodel filter=lfs diff=lfs merge=lfs -text
70
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l4_together.axmodel filter=lfs diff=lfs merge=lfs -text
71
+ video/frame_0040.jpg filter=lfs diff=lfs merge=lfs -text
72
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l0_together.axmodel filter=lfs diff=lfs merge=lfs -text
73
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l7_together.axmodel filter=lfs diff=lfs merge=lfs -text
74
+ video/frame_0016.jpg filter=lfs diff=lfs merge=lfs -text
75
+ video/frame_0048.jpg filter=lfs diff=lfs merge=lfs -text
76
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l19_together.axmodel filter=lfs diff=lfs merge=lfs -text
77
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l29_together.axmodel filter=lfs diff=lfs merge=lfs -text
78
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l9_together.axmodel filter=lfs diff=lfs merge=lfs -text
79
+ images/recoAll_attractions_4.jpg filter=lfs diff=lfs merge=lfs -text
80
+ images/recoAll_attractions_1.jpg filter=lfs diff=lfs merge=lfs -text
81
+ video/frame_0024.jpg filter=lfs diff=lfs merge=lfs -text
82
+ video/frame_0056.jpg filter=lfs diff=lfs merge=lfs -text
83
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l25_together.axmodel filter=lfs diff=lfs merge=lfs -text
84
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l26_together.axmodel filter=lfs diff=lfs merge=lfs -text
85
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l21_together.axmodel filter=lfs diff=lfs merge=lfs -text
86
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l31_together.axmodel filter=lfs diff=lfs merge=lfs -text
87
+ main_axcl_aarch64 filter=lfs diff=lfs merge=lfs -text
88
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l14_together.axmodel filter=lfs diff=lfs merge=lfs -text
89
+ images/recoAll_attractions_2.jpg filter=lfs diff=lfs merge=lfs -text
90
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l1_together.axmodel filter=lfs diff=lfs merge=lfs -text
91
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l23_together.axmodel filter=lfs diff=lfs merge=lfs -text
92
+ Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l32_together.axmodel filter=lfs diff=lfs merge=lfs -text
93
+ main_ax650 filter=lfs diff=lfs merge=lfs -text
Qwen3-VL-4B-Instruct-AX650-c128_p1152/Qwen3-VL-4B-Instruct_vision.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec5e5ce705711ae0660a0fa849c154219eb97288c4fc3fa9e55f6a2c5d1bd099
3
+ size 461423557
Qwen3-VL-4B-Instruct-AX650-c128_p1152/model.embed_tokens.weight.bfloat16.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2942e869a0df443edd3158b7f6c4f735f79faea563354b3a077db81f5f21edaa
3
+ size 777912320
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l0_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:724a7ab653edf8289533a476aa7b20fc671e3f6d71210030478f4470adf3bb64
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l10_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d491ba51dfc4619215b2aac13ee9e3a928bf44e4972abd8cf6071435b4219133
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l11_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4db0a97b819a751b11d9643382274be8400009eeb68e091850de27a4aa3450b
3
+ size 135516003
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l12_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0616fc5f016639db0da38ceae13d8cdea43f9665d77e5840ca1cd70d4a18a5a9
3
+ size 135516003
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l13_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c05eb728a25e20b8caec60276e12ae924ee329fab14a65a3b8db1bc55d8518ec
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l14_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f09a686295b1cfe06617520ce166965068dd0ee6365940b2e36807253e4ba8f4
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l15_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d9c4e5920fe30fc908d883099d5845256b92d7c7ff20ee3ec67509b22861800
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l16_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31914b23496641d399dace19e5e710c6b2ae396614ad57ed621175b8bc7cd89c
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l17_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccdc47138801e4ff174ed088c5acac66ccf33bc5ca08eefcc961af342201b511
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l18_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6eeaa31af70572139c57b67cad98a9abdd5d3541a149b5cd999f8439ad720473
3
+ size 135516067
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l19_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55120ba6bd45da9d4c3115ef1f1e4121107486c145135065026c2c5fa3351560
3
+ size 135515971
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l1_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6043dc1c76ee5b2d27cc9755b63ff6de80b8c3f8022aa7b51f5fe10e9baa39e9
3
+ size 135532035
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l20_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c870ffbdf45e579175a914458214eb6806b7bad7d7d74ea634f2ff2795a154f
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l21_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f595862558ab6ecd73c5df7e0ba29a8a7e13e2c50142354a9fb6b646f700f7c
3
+ size 135516003
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l22_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9cbb080e34bc5c4d3ba5da2f4093ccfe0af72d26df633eba8cede664ae64965c
3
+ size 135516835
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l23_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d16c565a821feeeb9a380dac4d09754de4b52a2a87cd45efba50800dfb7f1947
3
+ size 135515875
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l24_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ec3ef22f45a144789562616956109e72ebe42edaae96c0f099b54f26497e569
3
+ size 135516995
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l25_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9c1153b6bedac3db7bfebb50549702a1363e25eaca20513cece58b86cefc5ff
3
+ size 135516707
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l26_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69be1e3791e61f22a250490759e5069b37516ea3f14f3d9cd4a332fab6a28d74
3
+ size 135516515
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l27_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08ee124b07d1e43c98429a22ff17b954c6c69b81a24e732c9cbad4e5943be90a
3
+ size 135517379
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l28_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d82f725cb4926322043412ed21787f155d790cc8e757fbd9a4aa93dec6bf7d9a
3
+ size 135518083
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l29_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a0684f9bf0dca46e6c1c640efe73f7810a7ce8dc5a22adb3fbbb3f6f83a1c48
3
+ size 135516419
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l2_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bf8a93330ed8b41196991401f13c0cd66757bf640a4da2c3a79c27a01169a616
3
+ size 135521571
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l30_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:406e165a62c4d06c5bd9fe510713c4cb5cc7cbe2b592d617115506e09f45458f
3
+ size 135516643
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l31_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2271d6b0ef79f7573808c2c40d594f54ca0228892f4d270c88813570c608cf3b
3
+ size 135516387
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l32_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b66bffb786ca5a7c22c04d6c4e4616c58b4bd9216976518f82aa4945b50d3cd
3
+ size 135517379
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l33_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5e417f3a92827f5f13e350d464e528656f32bab278871063980bc254882895f
3
+ size 135516707
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l34_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b92a1dba0201662ce74c3db7fccc3070dff842e8ba5dc7551881ed983fb5d99
3
+ size 135516547
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l35_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ae2bf7a2f59cd01d115153aca4673b73fa2feb9019bf47339233d7578962536
3
+ size 135515971
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l3_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57f2f4e0e604f18d062f88dcc865a3c97e7247355f1eceb7a2ca74e8d71cf510
3
+ size 135529891
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l4_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecf0d111418f2ac101fc480515b896fca662389ac6cdc51fcedac9901f595342
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l5_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc69dee319165cfb62ed8e80646461b45115d0cbed5b62c589ceee99a51eb30d
3
+ size 135516771
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l6_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54b7d86e19eec4554ac2c48f539012029af8400b52a6c3fe96898282e5f1f1d9
3
+ size 135521571
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l7_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65eab95378f555225020d2d1a6cbb2d402ac6ca766f6ae10ca606bf68bd31cfe
3
+ size 135522115
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l8_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a8ec562fdf294f4e04ee958edb5c45fdfc33dee96966de12c866b475e5c1574
3
+ size 135515811
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_p128_l9_together.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d592ab3c972315bbb77d9073703742cbd0702d4bd6db7169c12c7c1f7b1a135
3
+ size 135516035
Qwen3-VL-4B-Instruct-AX650-c128_p1152/qwen3_vl_text_post.axmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:90df9d2cecd7ec005adc458b1432f690812bcfa6361654f654e42fef6664b0c7
3
+ size 424045076
README.md CHANGED
@@ -1,3 +1,233 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ language:
4
+ - en
5
+ - zh
6
+ base_model:
7
+ - Qwen/Qwen3-VL-2B-Instruct
8
+ - Qwen/Qwen3-VL-4B-Instruct
9
+ pipeline_tag: image-text-to-text
10
+ library_name: transformers
11
+ tags:
12
+ - Qwen3-VL
13
+ - Qwen3-VL-2B-Instruct
14
+ - Qwen3-VL-4B-Instruct
15
+ - Int8
16
+ - VLM
17
+ ---
18
+
19
+ # Qwen3-VL
20
+
21
+ This version of Qwen3-VL-2B-Instruct and Qwen3-VL-4B-Instruct have been converted to run on the Axera NPU using **w8a16** quantization.
22
+
23
+ Compatible with Pulsar2 version: 5.0
24
+
25
+ ## Convert tools links:
26
+
27
+ For those who are interested in model conversion, you can try to export axmodel through the original repo :
28
+
29
+ - https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct
30
+ - https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct
31
+
32
+ [Pulsar2 Link, How to Convert LLM from Huggingface to axmodel](https://pulsar2-docs.readthedocs.io/en/latest/appendix/build_llm.html)
33
+
34
+ [AXera NPU HOST LLM Runtime](https://github.com/AXERA-TECH/Qwen3-VL.AXERA)
35
+
36
+
37
+ ## Support Platform
38
+
39
+ - AX650
40
+ - AX650N DEMO Board
41
+ - [M4N-Dock(爱芯派Pro)](https://wiki.sipeed.com/hardware/zh/maixIV/m4ndock/m4ndock.html)
42
+ - [M.2 Accelerator card](https://axcl-docs.readthedocs.io/zh-cn/latest/doc_guide_hardware.html)
43
+
44
+ **Image Process**
45
+ |Chips| input size | image num | image encoder | ttft(320 tokens) | w8a16 |
46
+ |--|--|--|--|--|--|
47
+ |AX650| 384*384 | 1 | 207 ms | 392 ms | 9.5 tokens/sec|
48
+
49
+ **Video Process**
50
+ |Chips| input size | image num | image encoder |ttft(600 tokens) | w8a16 |
51
+ |--|--|--|--|--|--|
52
+ |AX650| 384*384 | 8 | 725 ms | 1045 ms | 9.5 tokens/sec|
53
+
54
+ The DDR capacity refers to the CMM memory that needs to be consumed. Ensure that the CMM memory allocation on the development board is greater than this value.
55
+
56
+ ## How to use
57
+
58
+ Download all files from this repository to the device
59
+
60
+ **If you using AX650 Board**
61
+
62
+ ### Prepare tokenizer server
63
+
64
+ #### Install transformer
65
+
66
+ ```
67
+ pip install -r requirements.txt
68
+ ```
69
+
70
+ ### Demo Run
71
+
72
+ #### Image understand demo
73
+
74
+ ##### start tokenizer server for image understand demo
75
+
76
+ ```
77
+ python3 tokenizer_images.py --port 8080
78
+ ```
79
+
80
+ ##### run image understand demo
81
+
82
+ - input text
83
+
84
+ ```
85
+ 描述这张图片
86
+ ```
87
+
88
+ - input image
89
+
90
+ ![](./images/demo.jpg)
91
+
92
+ ```
93
+ root@ax650 ~/Qwen3-VL # bash run_qwen3_vl_2b_image.sh
94
+ [I][ Init][ 156]: LLM init start
95
+ [I][ Init][ 34]: connect http://127.0.0.1:8080 ok
96
+ bos_id: -1, eos_id: 151645
97
+ img_start_token: 151652
98
+ img_context_token: 151655
99
+ 3% | ██ | 1 / 31 [0.01s<0.31s, 100.00 count/s] tokenizer init ok[I][ Init][ 26]: LLaMaEmbedSelector use mmap
100
+ 6% | ███ | 2 / 31 [0.02s<0.25s, 125.00 count/s] embed_selector init ok[I][ Init][ 198]: attr.axmodel_num:28
101
+ 103% | ██████████████████████████████████ | 32 / 31 [6.58s<6.37s, 4.86 count/s] init vpm axmodel ok,remain_cmm(3678 MB)[I][ Init][ 263]: IMAGE_CONTEXT_TOKEN: 151655, IMAGE_START_TOKEN: 151652
102
+ [I][ Init][ 306]: image encoder output float32
103
+
104
+ [I][ Init][ 336]: max_token_len : 2047
105
+ [I][ Init][ 341]: kv_cache_size : 1024, kv_cache_num: 2047
106
+ [I][ Init][ 349]: prefill_token_num : 128
107
+ [I][ Init][ 353]: grp: 1, prefill_max_token_num : 1
108
+ [I][ Init][ 353]: grp: 2, prefill_max_token_num : 128
109
+ [I][ Init][ 353]: grp: 3, prefill_max_token_num : 256
110
+ [I][ Init][ 353]: grp: 4, prefill_max_token_num : 384
111
+ [I][ Init][ 353]: grp: 5, prefill_max_token_num : 512
112
+ [I][ Init][ 353]: grp: 6, prefill_max_token_num : 640
113
+ [I][ Init][ 353]: grp: 7, prefill_max_token_num : 768
114
+ [I][ Init][ 353]: grp: 8, prefill_max_token_num : 896
115
+ [I][ Init][ 353]: grp: 9, prefill_max_token_num : 1024
116
+ [I][ Init][ 353]: grp: 10, prefill_max_token_num : 1152
117
+ [I][ Init][ 357]: prefill_max_token_num : 1152
118
+ [I][ Init][ 366]: LLM init ok
119
+ Type "q" to exit, Ctrl+c to stop current running
120
+ prompt >> 描述这张图片
121
+ image >> images/demo.jpg
122
+ [I][ Encode][ 490]: image encode time : 207.362000 ms, size : 1
123
+ [I][ Encode][ 533]: input_ids size:167
124
+ [I][ Encode][ 541]: offset 15
125
+ [I][ Encode][ 557]: img_embed.size:1, 294912
126
+ [I][ Encode][ 571]: out_embed size:342016
127
+ [I][ Encode][ 573]: position_ids size:167
128
+ [I][ Run][ 591]: input token num : 167, prefill_split_num : 2
129
+ [I][ Run][ 625]: input_num_token:128
130
+ [I][ Run][ 625]: input_num_token:39
131
+ [I][ Run][ 786]: ttft: 391.51 ms
132
+ 好的，这是一张描绘海滩上温馨场景的照片。
133
+
134
+ 照片捕捉了一个宁静而美好的瞬间：一位女士和她的狗在沙滩上。她们正坐在柔软的沙地上，背景是波光粼粼的海面和柔和的天空。阳光从画面的上方洒下，为整个场景镀上了一层温暖的金色，营造出一种宁静、舒适和幸福的氛围。
135
+
136
+ 女士穿着一件格子衬衫，她正侧身对着镜头，脸上带着温柔的微笑，似乎在与她的狗互动。她的狗，一只毛茸茸的浅色狗，正用前爪轻轻搭在她的腿上，显得非常亲密和信任。狗的尾巴微微翘起，显示出它兴奋和快乐的情绪。
137
+
138
+ 整个画面充满了爱与陪伴的温暖感觉，展现了人与宠物之间深厚的情感联系。
139
+
140
+ [N][ Run][ 913]: hit eos,avg 9.35 token/s
141
+
142
+ prompt >>
143
+ ```
144
+
145
+ #### Video understand demo
146
+
147
+ ##### start tokenizer server for image understand demo
148
+
149
+ ```
150
+ python tokenizer_video.py --port 8080
151
+ ```
152
+
153
+ ##### run video understand demo
154
+ - input text
155
+
156
+ ```
157
+ 描述这个视频
158
+ ```
159
+
160
+ - input video
161
+
162
+ ./video
163
+
164
+ ```
165
+ root@ax650 ~/Qwen3-VL # bash run_qwen3_vl_2b_video.sh
166
+ [I][ Init][ 156]: LLM init start
167
+ [I][ Init][ 34]: connect http://127.0.0.1:8080 ok
168
+ bos_id: -1, eos_id: 151645
169
+ img_start_token: 151652
170
+ img_context_token: 151656
171
+ 3% | ██ | 1 / 31 [0.01s<0.31s, 100.00 count/s] tokenizer init ok[I][ Init][ 26]: LLaMaEmbedSelector use mmap
172
+ 6% | ███ | 2 / 31 [0.01s<0.20s, 153.85 count/s] embed_selector init ok[I][ Init][ 198]: attr.axmodel_num:28
173
+ 103% | ██████████████████████████████████ | 32 / 31 [30.34s<29.39s, 1.05 count/s] init vpm axmodel ok,remain_cmm(3678 MB)[I][ Init][ 263]: IMAGE_CONTEXT_TOKEN: 151656, IMAGE_START_TOKEN: 151652
174
+ [I][ Init][ 306]: image encoder output float32
175
+
176
+ [I][ Init][ 336]: max_token_len : 2047
177
+ [I][ Init][ 341]: kv_cache_size : 1024, kv_cache_num: 2047
178
+ [I][ Init][ 349]: prefill_token_num : 128
179
+ [I][ Init][ 353]: grp: 1, prefill_max_token_num : 1
180
+ [I][ Init][ 353]: grp: 2, prefill_max_token_num : 128
181
+ [I][ Init][ 353]: grp: 3, prefill_max_token_num : 256
182
+ [I][ Init][ 353]: grp: 4, prefill_max_token_num : 384
183
+ [I][ Init][ 353]: grp: 5, prefill_max_token_num : 512
184
+ [I][ Init][ 353]: grp: 6, prefill_max_token_num : 640
185
+ [I][ Init][ 353]: grp: 7, prefill_max_token_num : 768
186
+ [I][ Init][ 353]: grp: 8, prefill_max_token_num : 896
187
+ [I][ Init][ 353]: grp: 9, prefill_max_token_num : 1024
188
+ [I][ Init][ 353]: grp: 10, prefill_max_token_num : 1152
189
+ [I][ Init][ 357]: prefill_max_token_num : 1152
190
+ [I][ Init][ 366]: LLM init ok
191
+ Type "q" to exit, Ctrl+c to stop current running
192
+ prompt >> 描述这个视频
193
+ image >> video
194
+ video/frame_0000.jpg
195
+ video/frame_0008.jpg
196
+ video/frame_0016.jpg
197
+ video/frame_0024.jpg
198
+ video/frame_0032.jpg
199
+ video/frame_0040.jpg
200
+ video/frame_0048.jpg
201
+ video/frame_0056.jpg
202
+ [I][ Encode][ 490]: image encode time : 751.804993 ms, size : 4
203
+ [I][ Encode][ 533]: input_ids size:600
204
+ [I][ Encode][ 541]: offset 15
205
+ [I][ Encode][ 557]: img_embed.size:4, 294912
206
+ [I][ Encode][ 562]: offset:159
207
+ [I][ Encode][ 562]: offset:303
208
+ [I][ Encode][ 562]: offset:447
209
+ [I][ Encode][ 571]: out_embed size:1228800
210
+ [I][ Encode][ 573]: position_ids size:600
211
+ [I][ Run][ 591]: input token num : 600, prefill_split_num : 5
212
+ [I][ Run][ 625]: input_num_token:128
213
+ [I][ Run][ 625]: input_num_token:128
214
+ [I][ Run][ 625]: input_num_token:128
215
+ [I][ Run][ 625]: input_num_token:128
216
+ [I][ Run][ 625]: input_num_token:88
217
+ [I][ Run][ 786]: ttft: 1040.91 ms
218
+ 根据您提供的图片，这是一段关于两只土拨鼠在山地环境中互动的视频片段。
219
+
220
+ - **主体**：画面中有两只土拨鼠（也称“山地土拨鼠”或“黑背土拨鼠”），它们正站在一块布满碎石的草地上。它们的毛色为灰褐色与黑色相间，面部有明显的黑色条纹，这是土拨鼠的典型特征。
221
+
222
+ - **行为**：这两只土拨鼠正进行着一种看似玩耍或社交的互动。它们用前爪互相拍打，身体前倾，姿态充满活力。这种行为在土拨鼠中通常表示友好、玩耍或建立社交联系。
223
+
224
+ - **环境**：背景是连绵起伏的山脉，山坡上覆盖着绿色的植被，天空晴朗，阳光明媚。整个场景给人一种自然、宁静又充满生机的感觉。
225
+
226
+ - **视频风格**：从画面的清晰度和动态感来看，这可能是一段慢动作或高清晰度的视频片段，捕捉了土拨鼠活泼、生动的瞬间。
227
+
228
+ 综上所述，这段视频生动地记录了两只土拨鼠在自然山地环境中友好互动的场景，展现了它们活泼、充满活力的天性。
229
+
230
+ [N][ Run][ 913]: hit eos,avg 9.44 token/s
231
+
232
+ prompt >>
233
+ ```
config.json ADDED
File without changes
images/demo.jpg ADDED

Git LFS Details

  • SHA256: 344d5f0e43bfd6a4a6ed655dc9c3dc76a2a6ecb1de3afeee998e61722378e8b3
  • Pointer size: 130 Bytes
  • Size of remote file: 64.5 kB
images/demo1.jpg ADDED

Git LFS Details

  • SHA256: 8d6156cbaa86cf0f4a9ebbfc8c2bf307dcf6fc64e32176d69333f6f7c7b294c5
  • Pointer size: 131 Bytes
  • Size of remote file: 101 kB
images/recoAll_attractions_1.jpg ADDED

Git LFS Details

  • SHA256: 0072430513e76580c4134b78e452a1fb729112fe5725d1f8481e697c7b5cd4a1
  • Pointer size: 130 Bytes
  • Size of remote file: 73.3 kB
images/recoAll_attractions_2.jpg ADDED

Git LFS Details

  • SHA256: 28acba6c284d06039ca57c9ce182094baafa8a48b49068dc44482a6d643689cc
  • Pointer size: 131 Bytes
  • Size of remote file: 105 kB
images/recoAll_attractions_3.jpg ADDED

Git LFS Details

  • SHA256: f308b3ebb1855b74768d9a24a256616f683ba59e52c238326501f4008a3ab9a4
  • Pointer size: 130 Bytes
  • Size of remote file: 59.4 kB
images/recoAll_attractions_4.jpg ADDED

Git LFS Details

  • SHA256: e47e13a69c16073e221b143a43263fcbe956bcfeb8ad47fddbdbc03a2b6f7261
  • Pointer size: 131 Bytes
  • Size of remote file: 115 kB
images/ssd_car.jpg ADDED

Git LFS Details

  • SHA256: 92d459a39a9eef03956257cf9fec84114d9e5df8fb9c0662fb257488cdd4f365
  • Pointer size: 130 Bytes
  • Size of remote file: 50.5 kB
images/ssd_horse.jpg ADDED

Git LFS Details

  • SHA256: ed22f6b4c8c33e50e391e089ede14e8fa9402c623b09dbcf010e804770698fbb
  • Pointer size: 131 Bytes
  • Size of remote file: 123 kB