llmmlman commited on
Commit
4a1c92a
·
verified ·
1 Parent(s): 1547a69

Upload folder using huggingface_hub

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 +2 -0
  2. ggml/weights.gguf +3 -0
  3. model.json +27 -0
  4. qnn/config.json +147 -0
  5. qnn/kv/layer_0_key_0.raw +0 -0
  6. qnn/kv/layer_0_key_1.raw +0 -0
  7. qnn/kv/layer_0_key_2.raw +0 -0
  8. qnn/kv/layer_0_key_3.raw +0 -0
  9. qnn/kv/layer_0_key_4.raw +0 -0
  10. qnn/kv/layer_0_key_5.raw +0 -0
  11. qnn/kv/layer_0_key_6.raw +0 -0
  12. qnn/kv/layer_0_key_7.raw +5 -0
  13. qnn/kv/layer_0_value_0.raw +0 -0
  14. qnn/kv/layer_0_value_1.raw +0 -0
  15. qnn/kv/layer_0_value_2.raw +2 -0
  16. qnn/kv/layer_0_value_3.raw +0 -0
  17. qnn/kv/layer_0_value_4.raw +0 -0
  18. qnn/kv/layer_0_value_5.raw +0 -0
  19. qnn/kv/layer_0_value_6.raw +0 -0
  20. qnn/kv/layer_0_value_7.raw +0 -0
  21. qnn/kv/layer_10_key_0.raw +0 -0
  22. qnn/kv/layer_10_key_1.raw +0 -0
  23. qnn/kv/layer_10_key_2.raw +0 -0
  24. qnn/kv/layer_10_key_3.raw +0 -0
  25. qnn/kv/layer_10_key_4.raw +0 -0
  26. qnn/kv/layer_10_key_5.raw +0 -0
  27. qnn/kv/layer_10_key_6.raw +1 -0
  28. qnn/kv/layer_10_key_7.raw +0 -0
  29. qnn/kv/layer_10_value_0.raw +0 -0
  30. qnn/kv/layer_10_value_1.raw +4 -0
  31. qnn/kv/layer_10_value_2.raw +0 -0
  32. qnn/kv/layer_10_value_3.raw +0 -0
  33. qnn/kv/layer_10_value_4.raw +2 -0
  34. qnn/kv/layer_10_value_5.raw +3 -0
  35. qnn/kv/layer_10_value_6.raw +3 -0
  36. qnn/kv/layer_10_value_7.raw +0 -0
  37. qnn/kv/layer_11_key_0.raw +0 -0
  38. qnn/kv/layer_11_key_1.raw +0 -0
  39. qnn/kv/layer_11_key_2.raw +0 -0
  40. qnn/kv/layer_11_key_3.raw +0 -0
  41. qnn/kv/layer_11_key_4.raw +0 -0
  42. qnn/kv/layer_11_key_5.raw +0 -0
  43. qnn/kv/layer_11_key_6.raw +4 -0
  44. qnn/kv/layer_11_key_7.raw +0 -0
  45. qnn/kv/layer_11_value_0.raw +3 -0
  46. qnn/kv/layer_11_value_1.raw +6 -0
  47. qnn/kv/layer_11_value_2.raw +0 -0
  48. qnn/kv/layer_11_value_3.raw +0 -0
  49. qnn/kv/layer_11_value_4.raw +0 -0
  50. qnn/kv/layer_11_value_5.raw +0 -0
.gitattributes CHANGED
@@ -33,3 +33,5 @@ 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
+ ggml/weights.gguf filter=lfs diff=lfs merge=lfs -text
37
+ vocab.gguf filter=lfs diff=lfs merge=lfs -text
ggml/weights.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d931447a4526cf4b2a952521f92d6acad4ee5585fb32291203b38d39eef0f6c
3
+ size 4525775456
model.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_id": "test",
3
+ "model_arch": "llama",
4
+ "version": 20241202,
5
+ "llm_config": {
6
+ "embed_dim": 4096,
7
+ "ffn_dim": 14336,
8
+ "head_size": 128,
9
+ "kv_dim": 1024,
10
+ "n_attn_heads": 32,
11
+ "n_attn_kv_heads": 8,
12
+ "n_ctx": 131072,
13
+ "n_layers": 32,
14
+ "norm_eps": 9.999999747378752e-06,
15
+ "vocab_size": 128256,
16
+ "rope_config": {
17
+ "n_rope_ctx_orig": 131072,
18
+ "rope_attn_factor": 1.0,
19
+ "rope_dim": 128,
20
+ "rope_freq_base": 500000.0,
21
+ "rope_freq_scale": 1.0,
22
+ "rope_scale_type": "linear",
23
+ "rope_type": 0
24
+ }
25
+ },
26
+ "vision": {}
27
+ }
qnn/config.json ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_parameters": {
3
+ "n_layers": 32,
4
+ "vocab_size": 128256,
5
+ "embed_dim": 4096,
6
+ "ffn_hidden_dim": 14336,
7
+ "head_dim": 128,
8
+ "n_kv_heads": 8,
9
+ "rope_theta": 500000.0,
10
+ "rms_norm_eps": 1e-05,
11
+ "attention_mask_value": -100000.0,
12
+ "tie_embedding": false
13
+ },
14
+ "qnn_parameters": {
15
+ "n_hvx_threads": 4
16
+ },
17
+ "graphs": [
18
+ {
19
+ "type": "transformers",
20
+ "start_layer_id": 0,
21
+ "end_layer_id": 8,
22
+ "batch_size": 1,
23
+ "cache_size": 3968,
24
+ "context_size": 4096,
25
+ "graph_name": "batch_1",
26
+ "model_path": "llama3_1_8b_0.bin",
27
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
28
+ "kv_size": 1,
29
+ "x_name": "x",
30
+ "out_name": "out"
31
+ },
32
+ {
33
+ "type": "transformers",
34
+ "start_layer_id": 8,
35
+ "end_layer_id": 16,
36
+ "batch_size": 1,
37
+ "cache_size": 3968,
38
+ "context_size": 4096,
39
+ "graph_name": "batch_1",
40
+ "model_path": "llama3_1_8b_1.bin",
41
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
42
+ "kv_size": 1,
43
+ "x_name": "x",
44
+ "out_name": "out"
45
+ },
46
+ {
47
+ "type": "transformers",
48
+ "start_layer_id": 16,
49
+ "end_layer_id": 24,
50
+ "batch_size": 1,
51
+ "cache_size": 3968,
52
+ "context_size": 4096,
53
+ "graph_name": "batch_1",
54
+ "model_path": "llama3_1_8b_2.bin",
55
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
56
+ "kv_size": 1,
57
+ "x_name": "x",
58
+ "out_name": "out"
59
+ },
60
+ {
61
+ "type": "transformers",
62
+ "start_layer_id": 24,
63
+ "end_layer_id": 32,
64
+ "batch_size": 1,
65
+ "cache_size": 3968,
66
+ "context_size": 4096,
67
+ "graph_name": "batch_1",
68
+ "model_path": "llama3_1_8b_3.bin",
69
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
70
+ "kv_size": 1,
71
+ "x_name": "x",
72
+ "out_name": "out"
73
+ },
74
+ {
75
+ "type": "transformers",
76
+ "start_layer_id": 0,
77
+ "end_layer_id": 8,
78
+ "batch_size": 128,
79
+ "cache_size": 3968,
80
+ "context_size": 4096,
81
+ "graph_name": "batch_128",
82
+ "model_path": "llama3_1_8b_0.bin",
83
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
84
+ "kv_size": 1,
85
+ "x_name": "x",
86
+ "out_name": "out"
87
+ },
88
+ {
89
+ "type": "transformers",
90
+ "start_layer_id": 8,
91
+ "end_layer_id": 16,
92
+ "batch_size": 128,
93
+ "cache_size": 3968,
94
+ "context_size": 4096,
95
+ "graph_name": "batch_128",
96
+ "model_path": "llama3_1_8b_1.bin",
97
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
98
+ "kv_size": 1,
99
+ "x_name": "x",
100
+ "out_name": "out"
101
+ },
102
+ {
103
+ "type": "transformers",
104
+ "start_layer_id": 16,
105
+ "end_layer_id": 24,
106
+ "batch_size": 128,
107
+ "cache_size": 3968,
108
+ "context_size": 4096,
109
+ "graph_name": "batch_128",
110
+ "model_path": "llama3_1_8b_2.bin",
111
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
112
+ "kv_size": 1,
113
+ "x_name": "x",
114
+ "out_name": "out"
115
+ },
116
+ {
117
+ "type": "transformers",
118
+ "start_layer_id": 24,
119
+ "end_layer_id": 32,
120
+ "batch_size": 128,
121
+ "cache_size": 3968,
122
+ "context_size": 4096,
123
+ "graph_name": "batch_128",
124
+ "model_path": "llama3_1_8b_3.bin",
125
+ "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
126
+ "kv_size": 1,
127
+ "x_name": "x",
128
+ "out_name": "out"
129
+ }
130
+ ],
131
+ "embeddings": [
132
+ {
133
+ "graph_name": "batch_1",
134
+ "model_path": "lm_head.bin",
135
+ "batch_size": 1,
136
+ "x_name": "x",
137
+ "out_name": "logits"
138
+ },
139
+ {
140
+ "graph_name": "batch_128",
141
+ "model_path": "lm_head.bin",
142
+ "batch_size": 128,
143
+ "x_name": "x",
144
+ "out_name": "logits"
145
+ }
146
+ ]
147
+ }
qnn/kv/layer_0_key_0.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_1.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_2.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_4.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_5.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_6.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_key_7.raw ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ Rɻ�5�;;�D;
2
+
3
+ <��ս�~�]D<�#�<�#��;�=�롼
4
+ �}��<G�㽌oP�WK��5���7v�w�=��=3z~�M���&>����V+���+>
5
+ tA�&��=�������F�.��]'���4>g�=��4�\�'>��N>Y�>�m>
qnn/kv/layer_0_value_0.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_1.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_2.raw ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ ��ɺ}��9O52;�E�:b396�L:DT;��Z�����eHL��D%:�g'9:�0m���9��;+u�:�˺�N��B��:b���V]�:�{:ײ��.a�.8g�y�5��![����;,�;:��f;���9
2
+ ��OZ�S[�vPB:Ux:D>R;����G��:�\�����9��>�O";�;����9�aX:����@9��Y
qnn/kv/layer_0_value_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_4.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_5.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_6.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_0_value_7.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_0.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_1.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_2.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_4.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_5.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_key_6.raw ADDED
@@ -0,0 +1 @@
 
 
1
+ �<�͐];��Q;��#����:��9���8�g ��?���l;��:9Qd���
qnn/kv/layer_10_key_7.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_value_0.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_value_1.raw ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ /�������s�<N��������I<́��F��U�;�S�;��FA�Yt�9͋);YO<!ZN;=M"�Z����%����HF�8�E^;�$�;e�; J�;v1;,H#������˻O�;�Y�;�ł����:v]��G:W���n�$��;]���J[�{�J�,Z";��ܻ�u��t�;��»��;-`��J�
2
+ ��㭺
3
+ ĕ��к�w��\戼էA��[�:S��84�;�;zX�;)�.<S��j㿹&�;Ϲ<TI�+�9.���4dG;��N;��:��j��� ����/p9�f�:t=��^�:_�̻� ;�;5ͺ�Z�`8>�Ż�҉��j�;��Z�:Ca��a�)�Τ��
4
+ ��%��ګ�:�ź;���C�;^��;WQ��K�:<X�����,�pT�<�� �3|E��<�;]�7;H�I��m�.��;��;J�;���sQ
qnn/kv/layer_10_value_2.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_value_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_10_value_4.raw ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ +̹�0��L<�~%�&������m��:�J ��5�<�Q�<R�(�ʞ��}���O
2
+ ��<;[꘺���Y3>������;տ��]���>�ͺ8�R<
qnn/kv/layer_10_value_5.raw ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ �T�P/�;��3<Y]���4�: sb< ��;ز�H"��;���h���O���)<X+D7�u���5:��;Wg��R��:Y��@�9$�Իa����
2
+
3
+ <`#Z<s�M�If�>��;M���{�7;�m��h��"�;�E�!X�<�ʻF��;� <0�����N�ƞ.��%�;?&��M� <2p��!�g:C\�:ZM�����2x<��8<.�;@#��{��;�-л��="�K�񇏼Ó�8vM�;��+�S��� �\��;Ę��@��;��L��T; B�:��`<YT=<"_�]̮;�"��Lẛ<><{��<���;�]��HM�;5h<k����+�)��C�˻��μ�(�;:����;bU<�I�M�;'��;����B<xL;� ��A�<z=�:�w�;�Nʼ
qnn/kv/layer_10_value_6.raw ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ É@��{�;}�@;
2
+ @2��a<�xS>���R �`���-"�;�0�;_�L<BC4�W��; �ѻF�\����;�c;������:H�&A�;�/�:w1!<.���І&����;�;kp���J�;pp{:��<`k;�];����󔺔�<0����鿻d!:�����
3
+ x�_�
qnn/kv/layer_10_value_7.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_0.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_1.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_2.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_4.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_5.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_key_6.raw ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ �B�9"K;�i�:Kn����:o���4�;��198��;D��Hy���H;tG@:Dq�:\��~��:�:��H :7�[:x�,�>Ԏ:Ҧh��d;�BI;c������Q%��å:P[:�C�;!�.:_m<����b�:?���;�U#;ſN;=�9<i��=90=Vk���<��#��U�<s$�<3��=�F�;}��<a�<�Ū;�T6<8�jR=�
2
+
3
+ ;�O;s��;`�:pY�����:'�"�Ż�:�C<y�ͻJl��y�I<� �+S��o�"Ͻ�j�����tg�<������<
4
+ J�Z�мie���w<Ls�s�)�>]�<|�<[������$=�G��8R<*��
qnn/kv/layer_11_key_7.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_value_0.raw ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ c�< ܵ<�,�;��2�m<,���h�%�����Uջ*_=:�����);<`��7� >;��67f\�; ɖ:�����꺶����Z=<��<;�r;4���^!���Yr������u�<�wW<�5�:G�9>�A�1<YD;d�k;�%��iJ��
2
+ ;�*<�IQ9��'�<�9O<���;�~D��nٻ��{�A��Ŭ�i]��*�(�6��;]@<�ݜ�a��[�&�_�:�� �fX׺��+<��e;�N�;rNb��*Ȼ`d=Q�A��r;4t'���>�D�^sy;Vt�!쾼�Ό�t�;�WC<��
3
+ ��]�#��-X�P�+<��};Ł캪g�;A=N<��r�5���:q���B<�:�;.7"{<R�9�q�)<N�ڻP#��K��:ƀ$;N��ʍ<�l#<A��;
qnn/kv/layer_11_value_1.raw ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ ��@8qW
2
+ <���j-����l<���;q�0< W���Wػ��<I¾:ɢ��n�;7�<<�9�.e�:Ed��V&7<�I�!e�<e�;֧<��;�"�<0�<%�޻�G��C��F�<<L�E;����1%�� ����;UT��/L��պ"I���G���r:���<�<}v$���<����9�;
3
+ �;�����wK�o�#��;�e��갻�jD��oV�L�*<�X��m�z<
4
+ Q<���;��9��$<�B�ő�<�'
5
+ ��t�;l�:<�(�%p/<k}<�0R�Dػ�FsL�EV�;��ʻ@�S; ӆ;�I˻��[�'����o �����߱�����ֻ
6
+ �c:c1�<�U���A;�k��<t<��J;
qnn/kv/layer_11_value_2.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_value_3.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_value_4.raw ADDED
Binary file (512 Bytes). View file
 
qnn/kv/layer_11_value_5.raw ADDED
Binary file (512 Bytes). View file