Hugging Face's logo

Xenova
/
OpenELM-450M-Instruct

Text Generation
Transformers.js
ONNX
openelm
conversational
custom_code
Model card Files
xet
Community
2
Xenova HF Staff commited on
Commit
b5dceb7
·
verified ·
1 Parent(s): e41fee4

Upload folder using huggingface_hub

Browse files
Files changed (10) hide show
  1. config.json +100 -0
  2. generation_config.json +6 -0
  3. onnx/model.onnx +3 -0
  4. onnx/model_bnb4.onnx +3 -0
  5. onnx/model_fp16.onnx +3 -0
  6. onnx/model_int8.onnx +3 -0
  7. onnx/model_q4.onnx +3 -0
  8. onnx/model_quantized.onnx +3 -0
  9. onnx/model_uint8.onnx +3 -0
  10. quantize_config.json +132 -0
config.json ADDED
@@ -0,0 +1,100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "apple/OpenELM-450M-Instruct",
3
+ "activation_fn_name": "swish",
4
+ "architectures": [
5
+ "OpenELMForCausalLM"
6
+ ],
7
+ "auto_map": {
8
+ "AutoConfig": "apple/OpenELM-450M-Instruct--configuration_openelm.OpenELMConfig",
9
+ "AutoModelForCausalLM": "apple/OpenELM-450M-Instruct--modeling_openelm.OpenELMForCausalLM"
10
+ },
11
+ "bos_token_id": 1,
12
+ "eos_token_id": 2,
13
+ "ffn_dim_divisor": 256,
14
+ "ffn_multipliers": [
15
+ 0.5,
16
+ 0.68,
17
+ 0.87,
18
+ 1.05,
19
+ 1.24,
20
+ 1.42,
21
+ 1.61,
22
+ 1.79,
23
+ 1.97,
24
+ 2.16,
25
+ 2.34,
26
+ 2.53,
27
+ 2.71,
28
+ 2.89,
29
+ 3.08,
30
+ 3.26,
31
+ 3.45,
32
+ 3.63,
33
+ 3.82,
34
+ 4.0
35
+ ],
36
+ "ffn_with_glu": true,
37
+ "head_dim": 64,
38
+ "initializer_range": 0.02,
39
+ "max_context_length": 2048,
40
+ "model_dim": 1536,
41
+ "model_type": "openelm",
42
+ "normalization_layer_name": "rms_norm",
43
+ "normalize_qk_projections": true,
44
+ "num_gqa_groups": 4,
45
+ "num_kv_heads": [
46
+ 3,
47
+ 3,
48
+ 3,
49
+ 4,
50
+ 4,
51
+ 4,
52
+ 4,
53
+ 4,
54
+ 4,
55
+ 4,
56
+ 5,
57
+ 5,
58
+ 5,
59
+ 5,
60
+ 5,
61
+ 5,
62
+ 6,
63
+ 6,
64
+ 6,
65
+ 6
66
+ ],
67
+ "num_query_heads": [
68
+ 12,
69
+ 12,
70
+ 12,
71
+ 16,
72
+ 16,
73
+ 16,
74
+ 16,
75
+ 16,
76
+ 16,
77
+ 16,
78
+ 20,
79
+ 20,
80
+ 20,
81
+ 20,
82
+ 20,
83
+ 20,
84
+ 24,
85
+ 24,
86
+ 24,
87
+ 24
88
+ ],
89
+ "num_transformer_layers": 20,
90
+ "qkv_multipliers": [
91
+ 0.5,
92
+ 1.0
93
+ ],
94
+ "rope_freq_constant": 10000,
95
+ "rope_max_length": 4096,
96
+ "share_input_output_layers": true,
97
+ "transformers_version": "4.38.2",
98
+ "use_cache": true,
99
+ "vocab_size": 32000
100
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.38.2"
6
+ }
onnx/model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70f0b0a1c938f6a379904c5a875bc3db170c67b4db0286b7233d5bcde2afbcdc
3
+ size 2032462018
onnx/model_bnb4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf5ad2f548ce6b1189614c21652e2ae8cc3fd7f002a08ee61e9ce13170b0048
3
+ size 461146769
onnx/model_fp16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:866def23c1fbd676407a229c1027750304622cc7bf736e8821cec8834ca88ab2
3
+ size 1019112984
onnx/model_int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa43693120de488220fb7f8c92edddde78448aa6f516d37783a63a5776d7512a
3
+ size 514121019
onnx/model_q4.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f064a5b45c1e3d32d2dc68284ca068056e2acbddb0fa8beb165ddb1807e75fe
3
+ size 489715705
onnx/model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa43693120de488220fb7f8c92edddde78448aa6f516d37783a63a5776d7512a
3
+ size 514121019
onnx/model_uint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ceebd65902448f1ed7c60fee531ffaad0039f34ad0512abc29e1e4d4aeab9826
3
+ size 514121063
quantize_config.json ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fp16": {},
3
+ "q8": {
4
+ "per_model_config": {
5
+ "model": {
6
+ "op_types": [
7
+ "Add",
8
+ "And",
9
+ "Cast",
10
+ "Concat",
11
+ "Constant",
12
+ "ConstantOfShape",
13
+ "Div",
14
+ "Equal",
15
+ "Expand",
16
+ "Gather",
17
+ "MatMul",
18
+ "Mul",
19
+ "Neg",
20
+ "Pow",
21
+ "Range",
22
+ "ReduceMean",
23
+ "Reshape",
24
+ "ScatterND",
25
+ "Shape",
26
+ "Sigmoid",
27
+ "Slice",
28
+ "Softmax",
29
+ "Split",
30
+ "Sqrt",
31
+ "Sub",
32
+ "Tile",
33
+ "Transpose",
34
+ "Unsqueeze",
35
+ "Where"
36
+ ],
37
+ "weight_type": "QInt8"
38
+ }
39
+ },
40
+ "per_channel": false,
41
+ "reduce_range": false
42
+ },
43
+ "int8": {
44
+ "per_model_config": {
45
+ "model": {
46
+ "op_types": [
47
+ "Add",
48
+ "And",
49
+ "Cast",
50
+ "Concat",
51
+ "Constant",
52
+ "ConstantOfShape",
53
+ "Div",
54
+ "Equal",
55
+ "Expand",
56
+ "Gather",
57
+ "MatMul",
58
+ "Mul",
59
+ "Neg",
60
+ "Pow",
61
+ "Range",
62
+ "ReduceMean",
63
+ "Reshape",
64
+ "ScatterND",
65
+ "Shape",
66
+ "Sigmoid",
67
+ "Slice",
68
+ "Softmax",
69
+ "Split",
70
+ "Sqrt",
71
+ "Sub",
72
+ "Tile",
73
+ "Transpose",
74
+ "Unsqueeze",
75
+ "Where"
76
+ ],
77
+ "weight_type": "QInt8"
78
+ }
79
+ },
80
+ "per_channel": false,
81
+ "reduce_range": false
82
+ },
83
+ "uint8": {
84
+ "per_model_config": {
85
+ "model": {
86
+ "op_types": [
87
+ "Add",
88
+ "And",
89
+ "Cast",
90
+ "Concat",
91
+ "Constant",
92
+ "ConstantOfShape",
93
+ "Div",
94
+ "Equal",
95
+ "Expand",
96
+ "Gather",
97
+ "MatMul",
98
+ "Mul",
99
+ "Neg",
100
+ "Pow",
101
+ "Range",
102
+ "ReduceMean",
103
+ "Reshape",
104
+ "ScatterND",
105
+ "Shape",
106
+ "Sigmoid",
107
+ "Slice",
108
+ "Softmax",
109
+ "Split",
110
+ "Sqrt",
111
+ "Sub",
112
+ "Tile",
113
+ "Transpose",
114
+ "Unsqueeze",
115
+ "Where"
116
+ ],
117
+ "weight_type": "QUInt8"
118
+ }
119
+ },
120
+ "per_channel": false,
121
+ "reduce_range": false
122
+ },
123
+ "q4": {
124
+ "block_size": 32,
125
+ "is_symmetric": true,
126
+ "accuracy_level": null
127
+ },
128
+ "bnb4": {
129
+ "block_size": 64,
130
+ "quant_type": 1
131
+ }
132
+ }