Upload 20 files
Browse files- ar128_cl4096_1_of_4_qnn_ctx.onnx +3 -0
- ar128_cl4096_2_of_4_qnn_ctx.onnx +3 -0
- ar128_cl4096_3_of_4_qnn_ctx.onnx +3 -0
- ar128_cl4096_4_of_4_qnn_ctx.onnx +3 -0
- ar1_cl4096_1_of_4_qnn_ctx.onnx +3 -0
- ar1_cl4096_2_of_4_qnn_ctx.onnx +3 -0
- ar1_cl4096_3_of_4_qnn_ctx.onnx +3 -0
- ar1_cl4096_4_of_4_qnn_ctx.onnx +3 -0
- dequantizer.onnx +3 -0
- genai_config.json +580 -0
- genai_config_removed_sliding_window.json +576 -0
- position-processor.onnx +3 -0
- quantizer.onnx +3 -0
- special_tokens_map.json +30 -0
- tokenizer.json +0 -0
- tokenizer_config.json +131 -0
- weight_sharing_model_1_of_4.serialized.json +0 -0
- weight_sharing_model_2_of_4.serialized.json +0 -0
- weight_sharing_model_3_of_4.serialized.json +0 -0
- weight_sharing_model_4_of_4.serialized.json +0 -0
ar128_cl4096_1_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6503cecc31f3d91e43af7a9d5c3ef1002486803ba8e34e8379102b0aebcdd45
|
| 3 |
+
size 2476
|
ar128_cl4096_2_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e254dc746fe3ebe48e2a8286f5315b158282d286e6d09181c0e43220dc81a61
|
| 3 |
+
size 2581
|
ar128_cl4096_3_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf49cd2f4b551a98f4fb283ae881c5dda50763867604b2e54c130d219737c759
|
| 3 |
+
size 2599
|
ar128_cl4096_4_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:42ce267cd86f959b49714e6d13b34a1e9a432f485c99cb7bffd079836bc44c1e
|
| 3 |
+
size 2542
|
ar1_cl4096_1_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3a51a43eefc266fce6c94e8ac91edf8ff0b0180f0d2b23b0c7c40b10dc1d0b38
|
| 3 |
+
size 2453
|
ar1_cl4096_2_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b763fe2465a3307534df1dec3d062ee4ca65ac4b6c3364cfadd9433260f64f2
|
| 3 |
+
size 2558
|
ar1_cl4096_3_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b2c2fc605449bc436302c7b6cc348099e5b426fcbd66a35d44e9690356a79e5
|
| 3 |
+
size 2576
|
ar1_cl4096_4_of_4_qnn_ctx.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d7c7c2b98e5deb6463268656be57e24c130ec21c4fc267e47723a39134c334a
|
| 3 |
+
size 2519
|
dequantizer.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9905a234f604f4f8496e517bedab0a342f40e6b6b79765bd7e22185aba3dcdfb
|
| 3 |
+
size 226
|
genai_config.json
ADDED
|
@@ -0,0 +1,580 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"context_length": 4096,
|
| 5 |
+
"decoder": {
|
| 6 |
+
"session_options": {
|
| 7 |
+
"log_id": "onnxruntime-genai",
|
| 8 |
+
"provider_options": [],
|
| 9 |
+
"log_severity_level": 0
|
| 10 |
+
},
|
| 11 |
+
"filename": "model.onnx",
|
| 12 |
+
"head_size": 96,
|
| 13 |
+
"hidden_size": 3072,
|
| 14 |
+
"inputs": {
|
| 15 |
+
"input_ids": "input_ids",
|
| 16 |
+
"attention_mask": "attention_mask_before_processor",
|
| 17 |
+
"position_ids": "position_ids",
|
| 18 |
+
"past_key_names": "past_key_%d_in",
|
| 19 |
+
"past_value_names": "past_value_%d_in"
|
| 20 |
+
},
|
| 21 |
+
"outputs": {
|
| 22 |
+
"logits": "logits_dequantized",
|
| 23 |
+
"present_key_names": "past_key_%d_out",
|
| 24 |
+
"present_value_names": "past_value_%d_out"
|
| 25 |
+
},
|
| 26 |
+
"num_attention_heads": 32,
|
| 27 |
+
"num_hidden_layers": 32,
|
| 28 |
+
"num_key_value_heads": 32,
|
| 29 |
+
"sliding_window": {
|
| 30 |
+
"window_size": 128,
|
| 31 |
+
"pad_value": 128
|
| 32 |
+
},
|
| 33 |
+
"pipeline": [
|
| 34 |
+
{
|
| 35 |
+
"position_processor": {
|
| 36 |
+
"filename": "position-processor.onnx",
|
| 37 |
+
"inputs": [
|
| 38 |
+
"attention_mask_before_processor",
|
| 39 |
+
"position_ids"
|
| 40 |
+
],
|
| 41 |
+
"outputs": [
|
| 42 |
+
"attention_mask_before_quantizer",
|
| 43 |
+
"position_ids_cos_before_quantizer",
|
| 44 |
+
"position_ids_sin_before_quantizer"
|
| 45 |
+
],
|
| 46 |
+
"session_options": {
|
| 47 |
+
"log_id": "onnxruntime-genai.position_processor",
|
| 48 |
+
"provider_options": [
|
| 49 |
+
{}
|
| 50 |
+
]
|
| 51 |
+
}
|
| 52 |
+
},
|
| 53 |
+
"quantizer": {
|
| 54 |
+
"filename": "quantizer.onnx",
|
| 55 |
+
"inputs": [
|
| 56 |
+
"attention_mask_before_quantizer",
|
| 57 |
+
"position_ids_cos_before_quantizer",
|
| 58 |
+
"position_ids_sin_before_quantizer"
|
| 59 |
+
],
|
| 60 |
+
"outputs": [
|
| 61 |
+
"attention_mask",
|
| 62 |
+
"position_ids_cos",
|
| 63 |
+
"position_ids_sin"
|
| 64 |
+
],
|
| 65 |
+
"session_options": {
|
| 66 |
+
"log_id": "onnxruntime-genai.quantizer",
|
| 67 |
+
"provider_options": [
|
| 68 |
+
{}
|
| 69 |
+
]
|
| 70 |
+
}
|
| 71 |
+
},
|
| 72 |
+
"prompt-processor-1": {
|
| 73 |
+
"filename": "ar128_cl4096_1_of_4_qnn_ctx.onnx",
|
| 74 |
+
"inputs": [
|
| 75 |
+
"input_ids",
|
| 76 |
+
"past_key_0_in",
|
| 77 |
+
"past_key_5_in",
|
| 78 |
+
"past_value_5_in",
|
| 79 |
+
"past_value_0_in",
|
| 80 |
+
"past_key_6_in",
|
| 81 |
+
"past_value_6_in",
|
| 82 |
+
"past_key_7_in",
|
| 83 |
+
"past_value_7_in",
|
| 84 |
+
"past_key_1_in",
|
| 85 |
+
"past_value_1_in",
|
| 86 |
+
"past_key_2_in",
|
| 87 |
+
"past_value_2_in",
|
| 88 |
+
"past_key_3_in",
|
| 89 |
+
"past_value_3_in",
|
| 90 |
+
"past_key_4_in",
|
| 91 |
+
"past_value_4_in",
|
| 92 |
+
"position_ids_cos",
|
| 93 |
+
"position_ids_sin",
|
| 94 |
+
"attention_mask"
|
| 95 |
+
],
|
| 96 |
+
"outputs": [
|
| 97 |
+
"past_value_0_out",
|
| 98 |
+
"past_key_0_out",
|
| 99 |
+
"past_value_1_out",
|
| 100 |
+
"past_key_1_out",
|
| 101 |
+
"past_value_2_out",
|
| 102 |
+
"past_key_2_out",
|
| 103 |
+
"past_value_3_out",
|
| 104 |
+
"past_key_3_out",
|
| 105 |
+
"past_value_4_out",
|
| 106 |
+
"past_key_4_out",
|
| 107 |
+
"past_value_5_out",
|
| 108 |
+
"past_key_5_out",
|
| 109 |
+
"past_value_6_out",
|
| 110 |
+
"past_key_6_out",
|
| 111 |
+
"past_value_7_out",
|
| 112 |
+
"past_key_7_out",
|
| 113 |
+
"_model_layers_7_Add_1_Add_output_0"
|
| 114 |
+
],
|
| 115 |
+
"session_options": {
|
| 116 |
+
"log_id": "onnxruntime-genai.pp1",
|
| 117 |
+
"provider_options": [
|
| 118 |
+
{
|
| 119 |
+
"qnn": {
|
| 120 |
+
"backend_path": "QnnHtp.dll",
|
| 121 |
+
"htp_performance_mode": "burst",
|
| 122 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 123 |
+
"qnn_context_priority": "high"
|
| 124 |
+
}
|
| 125 |
+
}
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"run_on_token_gen": false
|
| 129 |
+
},
|
| 130 |
+
"prompt-processor-2": {
|
| 131 |
+
"filename": "ar128_cl4096_2_of_4_qnn_ctx.onnx",
|
| 132 |
+
"inputs": [
|
| 133 |
+
"_model_layers_7_Add_1_Add_output_0",
|
| 134 |
+
"past_key_8_in",
|
| 135 |
+
"past_key_13_in",
|
| 136 |
+
"past_value_13_in",
|
| 137 |
+
"past_value_8_in",
|
| 138 |
+
"past_key_14_in",
|
| 139 |
+
"past_value_14_in",
|
| 140 |
+
"past_key_15_in",
|
| 141 |
+
"past_value_15_in",
|
| 142 |
+
"past_key_9_in",
|
| 143 |
+
"past_value_9_in",
|
| 144 |
+
"past_key_10_in",
|
| 145 |
+
"past_value_10_in",
|
| 146 |
+
"past_key_11_in",
|
| 147 |
+
"past_value_11_in",
|
| 148 |
+
"past_key_12_in",
|
| 149 |
+
"past_value_12_in",
|
| 150 |
+
"position_ids_cos",
|
| 151 |
+
"position_ids_sin",
|
| 152 |
+
"attention_mask"
|
| 153 |
+
],
|
| 154 |
+
"outputs": [
|
| 155 |
+
"past_value_8_out",
|
| 156 |
+
"past_key_8_out",
|
| 157 |
+
"past_value_9_out",
|
| 158 |
+
"past_key_9_out",
|
| 159 |
+
"past_value_10_out",
|
| 160 |
+
"past_key_10_out",
|
| 161 |
+
"past_value_11_out",
|
| 162 |
+
"past_key_11_out",
|
| 163 |
+
"past_value_12_out",
|
| 164 |
+
"past_key_12_out",
|
| 165 |
+
"past_value_13_out",
|
| 166 |
+
"past_key_13_out",
|
| 167 |
+
"past_value_14_out",
|
| 168 |
+
"past_key_14_out",
|
| 169 |
+
"past_value_15_out",
|
| 170 |
+
"past_key_15_out",
|
| 171 |
+
"_model_layers_15_Add_1_Add_output_0"
|
| 172 |
+
],
|
| 173 |
+
"session_options": {
|
| 174 |
+
"log_id": "onnxruntime-genai.pp2",
|
| 175 |
+
"provider_options": [
|
| 176 |
+
{
|
| 177 |
+
"qnn": {
|
| 178 |
+
"backend_path": "QnnHtp.dll",
|
| 179 |
+
"htp_performance_mode": "burst",
|
| 180 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 181 |
+
"qnn_context_priority": "high"
|
| 182 |
+
}
|
| 183 |
+
}
|
| 184 |
+
]
|
| 185 |
+
},
|
| 186 |
+
"run_on_token_gen": false
|
| 187 |
+
},
|
| 188 |
+
"prompt-processor-3": {
|
| 189 |
+
"filename": "ar128_cl4096_3_of_4_qnn_ctx.onnx",
|
| 190 |
+
"inputs": [
|
| 191 |
+
"_model_layers_15_Add_1_Add_output_0",
|
| 192 |
+
"past_key_16_in",
|
| 193 |
+
"past_key_21_in",
|
| 194 |
+
"past_value_21_in",
|
| 195 |
+
"past_value_16_in",
|
| 196 |
+
"past_key_22_in",
|
| 197 |
+
"past_value_22_in",
|
| 198 |
+
"past_key_23_in",
|
| 199 |
+
"past_value_23_in",
|
| 200 |
+
"past_key_17_in",
|
| 201 |
+
"past_value_17_in",
|
| 202 |
+
"past_key_18_in",
|
| 203 |
+
"past_value_18_in",
|
| 204 |
+
"past_key_19_in",
|
| 205 |
+
"past_value_19_in",
|
| 206 |
+
"past_key_20_in",
|
| 207 |
+
"past_value_20_in",
|
| 208 |
+
"position_ids_cos",
|
| 209 |
+
"position_ids_sin",
|
| 210 |
+
"attention_mask"
|
| 211 |
+
],
|
| 212 |
+
"outputs": [
|
| 213 |
+
"past_value_16_out",
|
| 214 |
+
"past_key_16_out",
|
| 215 |
+
"past_value_17_out",
|
| 216 |
+
"past_key_17_out",
|
| 217 |
+
"past_value_18_out",
|
| 218 |
+
"past_key_18_out",
|
| 219 |
+
"past_value_19_out",
|
| 220 |
+
"past_key_19_out",
|
| 221 |
+
"past_value_20_out",
|
| 222 |
+
"past_key_20_out",
|
| 223 |
+
"past_value_21_out",
|
| 224 |
+
"past_key_21_out",
|
| 225 |
+
"past_value_22_out",
|
| 226 |
+
"past_key_22_out",
|
| 227 |
+
"past_value_23_out",
|
| 228 |
+
"past_key_23_out",
|
| 229 |
+
"_model_layers_23_Add_1_Add_output_0"
|
| 230 |
+
],
|
| 231 |
+
"session_options": {
|
| 232 |
+
"log_id": "onnxruntime-genai.pp3",
|
| 233 |
+
"provider_options": [
|
| 234 |
+
{
|
| 235 |
+
"qnn": {
|
| 236 |
+
"backend_path": "QnnHtp.dll",
|
| 237 |
+
"htp_performance_mode": "burst",
|
| 238 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 239 |
+
"qnn_context_priority": "high"
|
| 240 |
+
}
|
| 241 |
+
}
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"run_on_token_gen": false
|
| 245 |
+
},
|
| 246 |
+
"prompt-processor-4": {
|
| 247 |
+
"filename": "ar128_cl4096_4_of_4_qnn_ctx.onnx",
|
| 248 |
+
"inputs": [
|
| 249 |
+
"_model_layers_23_Add_1_Add_output_0",
|
| 250 |
+
"past_key_24_in",
|
| 251 |
+
"past_key_29_in",
|
| 252 |
+
"past_value_29_in",
|
| 253 |
+
"past_value_24_in",
|
| 254 |
+
"past_key_30_in",
|
| 255 |
+
"past_value_30_in",
|
| 256 |
+
"past_key_31_in",
|
| 257 |
+
"past_value_31_in",
|
| 258 |
+
"past_key_25_in",
|
| 259 |
+
"past_value_25_in",
|
| 260 |
+
"past_key_26_in",
|
| 261 |
+
"past_value_26_in",
|
| 262 |
+
"past_key_27_in",
|
| 263 |
+
"past_value_27_in",
|
| 264 |
+
"past_key_28_in",
|
| 265 |
+
"past_value_28_in",
|
| 266 |
+
"position_ids_cos",
|
| 267 |
+
"position_ids_sin",
|
| 268 |
+
"attention_mask"
|
| 269 |
+
],
|
| 270 |
+
"outputs": [
|
| 271 |
+
"past_value_24_out",
|
| 272 |
+
"past_key_24_out",
|
| 273 |
+
"past_value_25_out",
|
| 274 |
+
"past_key_25_out",
|
| 275 |
+
"past_value_26_out",
|
| 276 |
+
"past_key_26_out",
|
| 277 |
+
"past_value_27_out",
|
| 278 |
+
"past_key_27_out",
|
| 279 |
+
"past_value_28_out",
|
| 280 |
+
"past_key_28_out",
|
| 281 |
+
"past_value_29_out",
|
| 282 |
+
"past_key_29_out",
|
| 283 |
+
"past_value_30_out",
|
| 284 |
+
"past_key_30_out",
|
| 285 |
+
"past_value_31_out",
|
| 286 |
+
"past_key_31_out",
|
| 287 |
+
"logits"
|
| 288 |
+
],
|
| 289 |
+
"session_options": {
|
| 290 |
+
"log_id": "onnxruntime-genai.pp4",
|
| 291 |
+
"provider_options": [
|
| 292 |
+
{
|
| 293 |
+
"qnn": {
|
| 294 |
+
"backend_path": "QnnHtp.dll",
|
| 295 |
+
"htp_performance_mode": "burst",
|
| 296 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 297 |
+
"qnn_context_priority": "high"
|
| 298 |
+
}
|
| 299 |
+
}
|
| 300 |
+
]
|
| 301 |
+
},
|
| 302 |
+
"run_on_token_gen": false
|
| 303 |
+
},
|
| 304 |
+
"token-generator-1": {
|
| 305 |
+
"filename": "ar1_cl4096_1_of_4_qnn_ctx.onnx",
|
| 306 |
+
"inputs": [
|
| 307 |
+
"input_ids",
|
| 308 |
+
"past_key_0_in",
|
| 309 |
+
"past_key_5_in",
|
| 310 |
+
"past_value_5_in",
|
| 311 |
+
"past_value_0_in",
|
| 312 |
+
"past_key_6_in",
|
| 313 |
+
"past_value_6_in",
|
| 314 |
+
"past_key_7_in",
|
| 315 |
+
"past_value_7_in",
|
| 316 |
+
"past_key_1_in",
|
| 317 |
+
"past_value_1_in",
|
| 318 |
+
"past_key_2_in",
|
| 319 |
+
"past_value_2_in",
|
| 320 |
+
"past_key_3_in",
|
| 321 |
+
"past_value_3_in",
|
| 322 |
+
"past_key_4_in",
|
| 323 |
+
"past_value_4_in",
|
| 324 |
+
"position_ids_cos",
|
| 325 |
+
"position_ids_sin",
|
| 326 |
+
"attention_mask"
|
| 327 |
+
],
|
| 328 |
+
"outputs": [
|
| 329 |
+
"past_value_0_out",
|
| 330 |
+
"past_key_0_out",
|
| 331 |
+
"past_value_1_out",
|
| 332 |
+
"past_key_1_out",
|
| 333 |
+
"past_value_2_out",
|
| 334 |
+
"past_key_2_out",
|
| 335 |
+
"past_value_3_out",
|
| 336 |
+
"past_key_3_out",
|
| 337 |
+
"past_value_4_out",
|
| 338 |
+
"past_key_4_out",
|
| 339 |
+
"past_value_5_out",
|
| 340 |
+
"past_key_5_out",
|
| 341 |
+
"past_value_6_out",
|
| 342 |
+
"past_key_6_out",
|
| 343 |
+
"past_value_7_out",
|
| 344 |
+
"past_key_7_out",
|
| 345 |
+
"_model_layers_7_Add_1_Add_output_0"
|
| 346 |
+
],
|
| 347 |
+
"session_options": {
|
| 348 |
+
"log_id": "onnxruntime-genai.tg1",
|
| 349 |
+
"provider_options": [
|
| 350 |
+
{
|
| 351 |
+
"qnn": {
|
| 352 |
+
"backend_path": "QnnHtp.dll",
|
| 353 |
+
"htp_performance_mode": "burst",
|
| 354 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 355 |
+
"qnn_context_priority": "high"
|
| 356 |
+
}
|
| 357 |
+
}
|
| 358 |
+
]
|
| 359 |
+
},
|
| 360 |
+
"run_on_prompt": false
|
| 361 |
+
},
|
| 362 |
+
"token-generator-2": {
|
| 363 |
+
"filename": "ar1_cl4096_2_of_4_qnn_ctx.onnx",
|
| 364 |
+
"inputs": [
|
| 365 |
+
"_model_layers_7_Add_1_Add_output_0",
|
| 366 |
+
"past_key_8_in",
|
| 367 |
+
"past_key_13_in",
|
| 368 |
+
"past_value_13_in",
|
| 369 |
+
"past_value_8_in",
|
| 370 |
+
"past_key_14_in",
|
| 371 |
+
"past_value_14_in",
|
| 372 |
+
"past_key_15_in",
|
| 373 |
+
"past_value_15_in",
|
| 374 |
+
"past_key_9_in",
|
| 375 |
+
"past_value_9_in",
|
| 376 |
+
"past_key_10_in",
|
| 377 |
+
"past_value_10_in",
|
| 378 |
+
"past_key_11_in",
|
| 379 |
+
"past_value_11_in",
|
| 380 |
+
"past_key_12_in",
|
| 381 |
+
"past_value_12_in",
|
| 382 |
+
"position_ids_cos",
|
| 383 |
+
"position_ids_sin",
|
| 384 |
+
"attention_mask"
|
| 385 |
+
],
|
| 386 |
+
"outputs": [
|
| 387 |
+
"past_value_8_out",
|
| 388 |
+
"past_key_8_out",
|
| 389 |
+
"past_value_9_out",
|
| 390 |
+
"past_key_9_out",
|
| 391 |
+
"past_value_10_out",
|
| 392 |
+
"past_key_10_out",
|
| 393 |
+
"past_value_11_out",
|
| 394 |
+
"past_key_11_out",
|
| 395 |
+
"past_value_12_out",
|
| 396 |
+
"past_key_12_out",
|
| 397 |
+
"past_value_13_out",
|
| 398 |
+
"past_key_13_out",
|
| 399 |
+
"past_value_14_out",
|
| 400 |
+
"past_key_14_out",
|
| 401 |
+
"past_value_15_out",
|
| 402 |
+
"past_key_15_out",
|
| 403 |
+
"_model_layers_15_Add_1_Add_output_0"
|
| 404 |
+
],
|
| 405 |
+
"session_options": {
|
| 406 |
+
"log_id": "onnxruntime-genai.tg2",
|
| 407 |
+
"provider_options": [
|
| 408 |
+
{
|
| 409 |
+
"qnn": {
|
| 410 |
+
"backend_path": "QnnHtp.dll",
|
| 411 |
+
"htp_performance_mode": "burst",
|
| 412 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 413 |
+
"qnn_context_priority": "high"
|
| 414 |
+
}
|
| 415 |
+
}
|
| 416 |
+
]
|
| 417 |
+
},
|
| 418 |
+
"run_on_prompt": false
|
| 419 |
+
},
|
| 420 |
+
"token-generator-3": {
|
| 421 |
+
"filename": "ar1_cl4096_3_of_4_qnn_ctx.onnx",
|
| 422 |
+
"inputs": [
|
| 423 |
+
"_model_layers_15_Add_1_Add_output_0",
|
| 424 |
+
"past_key_16_in",
|
| 425 |
+
"past_key_21_in",
|
| 426 |
+
"past_value_21_in",
|
| 427 |
+
"past_value_16_in",
|
| 428 |
+
"past_key_22_in",
|
| 429 |
+
"past_value_22_in",
|
| 430 |
+
"past_key_23_in",
|
| 431 |
+
"past_value_23_in",
|
| 432 |
+
"past_key_17_in",
|
| 433 |
+
"past_value_17_in",
|
| 434 |
+
"past_key_18_in",
|
| 435 |
+
"past_value_18_in",
|
| 436 |
+
"past_key_19_in",
|
| 437 |
+
"past_value_19_in",
|
| 438 |
+
"past_key_20_in",
|
| 439 |
+
"past_value_20_in",
|
| 440 |
+
"position_ids_cos",
|
| 441 |
+
"position_ids_sin",
|
| 442 |
+
"attention_mask"
|
| 443 |
+
],
|
| 444 |
+
"outputs": [
|
| 445 |
+
"past_value_16_out",
|
| 446 |
+
"past_key_16_out",
|
| 447 |
+
"past_value_17_out",
|
| 448 |
+
"past_key_17_out",
|
| 449 |
+
"past_value_18_out",
|
| 450 |
+
"past_key_18_out",
|
| 451 |
+
"past_value_19_out",
|
| 452 |
+
"past_key_19_out",
|
| 453 |
+
"past_value_20_out",
|
| 454 |
+
"past_key_20_out",
|
| 455 |
+
"past_value_21_out",
|
| 456 |
+
"past_key_21_out",
|
| 457 |
+
"past_value_22_out",
|
| 458 |
+
"past_key_22_out",
|
| 459 |
+
"past_value_23_out",
|
| 460 |
+
"past_key_23_out",
|
| 461 |
+
"_model_layers_23_Add_1_Add_output_0"
|
| 462 |
+
],
|
| 463 |
+
"session_options": {
|
| 464 |
+
"log_id": "onnxruntime-genai.tg3",
|
| 465 |
+
"provider_options": [
|
| 466 |
+
{
|
| 467 |
+
"qnn": {
|
| 468 |
+
"backend_path": "QnnHtp.dll",
|
| 469 |
+
"htp_performance_mode": "burst",
|
| 470 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 471 |
+
"qnn_context_priority": "high"
|
| 472 |
+
}
|
| 473 |
+
}
|
| 474 |
+
]
|
| 475 |
+
},
|
| 476 |
+
"run_on_prompt": false
|
| 477 |
+
},
|
| 478 |
+
"token-generator-4": {
|
| 479 |
+
"filename": "ar1_cl4096_4_of_4_qnn_ctx.onnx",
|
| 480 |
+
"inputs": [
|
| 481 |
+
"_model_layers_23_Add_1_Add_output_0",
|
| 482 |
+
"past_key_24_in",
|
| 483 |
+
"past_key_29_in",
|
| 484 |
+
"past_value_29_in",
|
| 485 |
+
"past_value_24_in",
|
| 486 |
+
"past_key_30_in",
|
| 487 |
+
"past_value_30_in",
|
| 488 |
+
"past_key_31_in",
|
| 489 |
+
"past_value_31_in",
|
| 490 |
+
"past_key_25_in",
|
| 491 |
+
"past_value_25_in",
|
| 492 |
+
"past_key_26_in",
|
| 493 |
+
"past_value_26_in",
|
| 494 |
+
"past_key_27_in",
|
| 495 |
+
"past_value_27_in",
|
| 496 |
+
"past_key_28_in",
|
| 497 |
+
"past_value_28_in",
|
| 498 |
+
"position_ids_cos",
|
| 499 |
+
"position_ids_sin",
|
| 500 |
+
"attention_mask"
|
| 501 |
+
],
|
| 502 |
+
"outputs": [
|
| 503 |
+
"past_value_24_out",
|
| 504 |
+
"past_key_24_out",
|
| 505 |
+
"past_value_25_out",
|
| 506 |
+
"past_key_25_out",
|
| 507 |
+
"past_value_26_out",
|
| 508 |
+
"past_key_26_out",
|
| 509 |
+
"past_value_27_out",
|
| 510 |
+
"past_key_27_out",
|
| 511 |
+
"past_value_28_out",
|
| 512 |
+
"past_key_28_out",
|
| 513 |
+
"past_value_29_out",
|
| 514 |
+
"past_key_29_out",
|
| 515 |
+
"past_value_30_out",
|
| 516 |
+
"past_key_30_out",
|
| 517 |
+
"past_value_31_out",
|
| 518 |
+
"past_key_31_out",
|
| 519 |
+
"logits"
|
| 520 |
+
],
|
| 521 |
+
"session_options": {
|
| 522 |
+
"log_id": "onnxruntime-genai.tg4",
|
| 523 |
+
"provider_options": [
|
| 524 |
+
{
|
| 525 |
+
"qnn": {
|
| 526 |
+
"backend_path": "QnnHtp.dll",
|
| 527 |
+
"htp_performance_mode": "burst",
|
| 528 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 529 |
+
"qnn_context_priority": "high"
|
| 530 |
+
}
|
| 531 |
+
}
|
| 532 |
+
]
|
| 533 |
+
},
|
| 534 |
+
"run_on_prompt": false
|
| 535 |
+
},
|
| 536 |
+
"dequantizer": {
|
| 537 |
+
"filename": "dequantizer.onnx",
|
| 538 |
+
"inputs": [
|
| 539 |
+
"logits"
|
| 540 |
+
],
|
| 541 |
+
"outputs": [
|
| 542 |
+
"logits_dequantized"
|
| 543 |
+
],
|
| 544 |
+
"session_options": {
|
| 545 |
+
"log_id": "onnxruntime-genai.dequantizer",
|
| 546 |
+
"provider_options": [
|
| 547 |
+
{}
|
| 548 |
+
]
|
| 549 |
+
}
|
| 550 |
+
}
|
| 551 |
+
}
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"eos_token_id": [
|
| 555 |
+
32007,
|
| 556 |
+
32001,
|
| 557 |
+
32000,
|
| 558 |
+
2
|
| 559 |
+
],
|
| 560 |
+
"pad_token_id": 32000,
|
| 561 |
+
"type": "decoder-pipeline",
|
| 562 |
+
"vocab_size": 32064
|
| 563 |
+
},
|
| 564 |
+
"search": {
|
| 565 |
+
"diversity_penalty": 0.0,
|
| 566 |
+
"do_sample": true,
|
| 567 |
+
"early_stopping": true,
|
| 568 |
+
"length_penalty": 1.0,
|
| 569 |
+
"max_length": 2048,
|
| 570 |
+
"min_length": 0,
|
| 571 |
+
"no_repeat_ngram_size": 0,
|
| 572 |
+
"num_beams": 1,
|
| 573 |
+
"num_return_sequences": 1,
|
| 574 |
+
"past_present_share_buffer": true,
|
| 575 |
+
"repetition_penalty": 1.0,
|
| 576 |
+
"temperature": 0.6,
|
| 577 |
+
"top_k": 1,
|
| 578 |
+
"top_p": 1.0
|
| 579 |
+
}
|
| 580 |
+
}
|
genai_config_removed_sliding_window.json
ADDED
|
@@ -0,0 +1,576 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"context_length": 4096,
|
| 5 |
+
"decoder": {
|
| 6 |
+
"session_options": {
|
| 7 |
+
"log_id": "onnxruntime-genai",
|
| 8 |
+
"provider_options": [],
|
| 9 |
+
"log_severity_level": 0
|
| 10 |
+
},
|
| 11 |
+
"filename": "model.onnx",
|
| 12 |
+
"head_size": 96,
|
| 13 |
+
"hidden_size": 3072,
|
| 14 |
+
"inputs": {
|
| 15 |
+
"input_ids": "input_ids",
|
| 16 |
+
"attention_mask": "attention_mask_before_processor",
|
| 17 |
+
"position_ids": "position_ids",
|
| 18 |
+
"past_key_names": "past_key_%d_in",
|
| 19 |
+
"past_value_names": "past_value_%d_in"
|
| 20 |
+
},
|
| 21 |
+
"outputs": {
|
| 22 |
+
"logits": "logits_dequantized",
|
| 23 |
+
"present_key_names": "past_key_%d_out",
|
| 24 |
+
"present_value_names": "past_value_%d_out"
|
| 25 |
+
},
|
| 26 |
+
"num_attention_heads": 32,
|
| 27 |
+
"num_hidden_layers": 32,
|
| 28 |
+
"num_key_value_heads": 32,
|
| 29 |
+
"pipeline": [
|
| 30 |
+
{
|
| 31 |
+
"position_processor": {
|
| 32 |
+
"filename": "position-processor.onnx",
|
| 33 |
+
"inputs": [
|
| 34 |
+
"attention_mask_before_processor",
|
| 35 |
+
"position_ids"
|
| 36 |
+
],
|
| 37 |
+
"outputs": [
|
| 38 |
+
"attention_mask_before_quantizer",
|
| 39 |
+
"position_ids_cos_before_quantizer",
|
| 40 |
+
"position_ids_sin_before_quantizer"
|
| 41 |
+
],
|
| 42 |
+
"session_options": {
|
| 43 |
+
"log_id": "onnxruntime-genai.position_processor",
|
| 44 |
+
"provider_options": [
|
| 45 |
+
{}
|
| 46 |
+
]
|
| 47 |
+
}
|
| 48 |
+
},
|
| 49 |
+
"quantizer": {
|
| 50 |
+
"filename": "quantizer.onnx",
|
| 51 |
+
"inputs": [
|
| 52 |
+
"attention_mask_before_quantizer",
|
| 53 |
+
"position_ids_cos_before_quantizer",
|
| 54 |
+
"position_ids_sin_before_quantizer"
|
| 55 |
+
],
|
| 56 |
+
"outputs": [
|
| 57 |
+
"attention_mask",
|
| 58 |
+
"position_ids_cos",
|
| 59 |
+
"position_ids_sin"
|
| 60 |
+
],
|
| 61 |
+
"session_options": {
|
| 62 |
+
"log_id": "onnxruntime-genai.quantizer",
|
| 63 |
+
"provider_options": [
|
| 64 |
+
{}
|
| 65 |
+
]
|
| 66 |
+
}
|
| 67 |
+
},
|
| 68 |
+
"prompt-processor-1": {
|
| 69 |
+
"filename": "ar128_cl4096_1_of_4_qnn_ctx.onnx",
|
| 70 |
+
"inputs": [
|
| 71 |
+
"input_ids",
|
| 72 |
+
"past_key_0_in",
|
| 73 |
+
"past_key_5_in",
|
| 74 |
+
"past_value_5_in",
|
| 75 |
+
"past_value_0_in",
|
| 76 |
+
"past_key_6_in",
|
| 77 |
+
"past_value_6_in",
|
| 78 |
+
"past_key_7_in",
|
| 79 |
+
"past_value_7_in",
|
| 80 |
+
"past_key_1_in",
|
| 81 |
+
"past_value_1_in",
|
| 82 |
+
"past_key_2_in",
|
| 83 |
+
"past_value_2_in",
|
| 84 |
+
"past_key_3_in",
|
| 85 |
+
"past_value_3_in",
|
| 86 |
+
"past_key_4_in",
|
| 87 |
+
"past_value_4_in",
|
| 88 |
+
"position_ids_cos",
|
| 89 |
+
"position_ids_sin",
|
| 90 |
+
"attention_mask"
|
| 91 |
+
],
|
| 92 |
+
"outputs": [
|
| 93 |
+
"past_value_0_out",
|
| 94 |
+
"past_key_0_out",
|
| 95 |
+
"past_value_1_out",
|
| 96 |
+
"past_key_1_out",
|
| 97 |
+
"past_value_2_out",
|
| 98 |
+
"past_key_2_out",
|
| 99 |
+
"past_value_3_out",
|
| 100 |
+
"past_key_3_out",
|
| 101 |
+
"past_value_4_out",
|
| 102 |
+
"past_key_4_out",
|
| 103 |
+
"past_value_5_out",
|
| 104 |
+
"past_key_5_out",
|
| 105 |
+
"past_value_6_out",
|
| 106 |
+
"past_key_6_out",
|
| 107 |
+
"past_value_7_out",
|
| 108 |
+
"past_key_7_out",
|
| 109 |
+
"_model_layers_7_Add_1_Add_output_0"
|
| 110 |
+
],
|
| 111 |
+
"session_options": {
|
| 112 |
+
"log_id": "onnxruntime-genai.pp1",
|
| 113 |
+
"provider_options": [
|
| 114 |
+
{
|
| 115 |
+
"qnn": {
|
| 116 |
+
"backend_path": "QnnHtp.dll",
|
| 117 |
+
"htp_performance_mode": "burst",
|
| 118 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 119 |
+
"qnn_context_priority": "high"
|
| 120 |
+
}
|
| 121 |
+
}
|
| 122 |
+
]
|
| 123 |
+
},
|
| 124 |
+
"run_on_token_gen": false
|
| 125 |
+
},
|
| 126 |
+
"prompt-processor-2": {
|
| 127 |
+
"filename": "ar128_cl4096_2_of_4_qnn_ctx.onnx",
|
| 128 |
+
"inputs": [
|
| 129 |
+
"_model_layers_7_Add_1_Add_output_0",
|
| 130 |
+
"past_key_8_in",
|
| 131 |
+
"past_key_13_in",
|
| 132 |
+
"past_value_13_in",
|
| 133 |
+
"past_value_8_in",
|
| 134 |
+
"past_key_14_in",
|
| 135 |
+
"past_value_14_in",
|
| 136 |
+
"past_key_15_in",
|
| 137 |
+
"past_value_15_in",
|
| 138 |
+
"past_key_9_in",
|
| 139 |
+
"past_value_9_in",
|
| 140 |
+
"past_key_10_in",
|
| 141 |
+
"past_value_10_in",
|
| 142 |
+
"past_key_11_in",
|
| 143 |
+
"past_value_11_in",
|
| 144 |
+
"past_key_12_in",
|
| 145 |
+
"past_value_12_in",
|
| 146 |
+
"position_ids_cos",
|
| 147 |
+
"position_ids_sin",
|
| 148 |
+
"attention_mask"
|
| 149 |
+
],
|
| 150 |
+
"outputs": [
|
| 151 |
+
"past_value_8_out",
|
| 152 |
+
"past_key_8_out",
|
| 153 |
+
"past_value_9_out",
|
| 154 |
+
"past_key_9_out",
|
| 155 |
+
"past_value_10_out",
|
| 156 |
+
"past_key_10_out",
|
| 157 |
+
"past_value_11_out",
|
| 158 |
+
"past_key_11_out",
|
| 159 |
+
"past_value_12_out",
|
| 160 |
+
"past_key_12_out",
|
| 161 |
+
"past_value_13_out",
|
| 162 |
+
"past_key_13_out",
|
| 163 |
+
"past_value_14_out",
|
| 164 |
+
"past_key_14_out",
|
| 165 |
+
"past_value_15_out",
|
| 166 |
+
"past_key_15_out",
|
| 167 |
+
"_model_layers_15_Add_1_Add_output_0"
|
| 168 |
+
],
|
| 169 |
+
"session_options": {
|
| 170 |
+
"log_id": "onnxruntime-genai.pp2",
|
| 171 |
+
"provider_options": [
|
| 172 |
+
{
|
| 173 |
+
"qnn": {
|
| 174 |
+
"backend_path": "QnnHtp.dll",
|
| 175 |
+
"htp_performance_mode": "burst",
|
| 176 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 177 |
+
"qnn_context_priority": "high"
|
| 178 |
+
}
|
| 179 |
+
}
|
| 180 |
+
]
|
| 181 |
+
},
|
| 182 |
+
"run_on_token_gen": false
|
| 183 |
+
},
|
| 184 |
+
"prompt-processor-3": {
|
| 185 |
+
"filename": "ar128_cl4096_3_of_4_qnn_ctx.onnx",
|
| 186 |
+
"inputs": [
|
| 187 |
+
"_model_layers_15_Add_1_Add_output_0",
|
| 188 |
+
"past_key_16_in",
|
| 189 |
+
"past_key_21_in",
|
| 190 |
+
"past_value_21_in",
|
| 191 |
+
"past_value_16_in",
|
| 192 |
+
"past_key_22_in",
|
| 193 |
+
"past_value_22_in",
|
| 194 |
+
"past_key_23_in",
|
| 195 |
+
"past_value_23_in",
|
| 196 |
+
"past_key_17_in",
|
| 197 |
+
"past_value_17_in",
|
| 198 |
+
"past_key_18_in",
|
| 199 |
+
"past_value_18_in",
|
| 200 |
+
"past_key_19_in",
|
| 201 |
+
"past_value_19_in",
|
| 202 |
+
"past_key_20_in",
|
| 203 |
+
"past_value_20_in",
|
| 204 |
+
"position_ids_cos",
|
| 205 |
+
"position_ids_sin",
|
| 206 |
+
"attention_mask"
|
| 207 |
+
],
|
| 208 |
+
"outputs": [
|
| 209 |
+
"past_value_16_out",
|
| 210 |
+
"past_key_16_out",
|
| 211 |
+
"past_value_17_out",
|
| 212 |
+
"past_key_17_out",
|
| 213 |
+
"past_value_18_out",
|
| 214 |
+
"past_key_18_out",
|
| 215 |
+
"past_value_19_out",
|
| 216 |
+
"past_key_19_out",
|
| 217 |
+
"past_value_20_out",
|
| 218 |
+
"past_key_20_out",
|
| 219 |
+
"past_value_21_out",
|
| 220 |
+
"past_key_21_out",
|
| 221 |
+
"past_value_22_out",
|
| 222 |
+
"past_key_22_out",
|
| 223 |
+
"past_value_23_out",
|
| 224 |
+
"past_key_23_out",
|
| 225 |
+
"_model_layers_23_Add_1_Add_output_0"
|
| 226 |
+
],
|
| 227 |
+
"session_options": {
|
| 228 |
+
"log_id": "onnxruntime-genai.pp3",
|
| 229 |
+
"provider_options": [
|
| 230 |
+
{
|
| 231 |
+
"qnn": {
|
| 232 |
+
"backend_path": "QnnHtp.dll",
|
| 233 |
+
"htp_performance_mode": "burst",
|
| 234 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 235 |
+
"qnn_context_priority": "high"
|
| 236 |
+
}
|
| 237 |
+
}
|
| 238 |
+
]
|
| 239 |
+
},
|
| 240 |
+
"run_on_token_gen": false
|
| 241 |
+
},
|
| 242 |
+
"prompt-processor-4": {
|
| 243 |
+
"filename": "ar128_cl4096_4_of_4_qnn_ctx.onnx",
|
| 244 |
+
"inputs": [
|
| 245 |
+
"_model_layers_23_Add_1_Add_output_0",
|
| 246 |
+
"past_key_24_in",
|
| 247 |
+
"past_key_29_in",
|
| 248 |
+
"past_value_29_in",
|
| 249 |
+
"past_value_24_in",
|
| 250 |
+
"past_key_30_in",
|
| 251 |
+
"past_value_30_in",
|
| 252 |
+
"past_key_31_in",
|
| 253 |
+
"past_value_31_in",
|
| 254 |
+
"past_key_25_in",
|
| 255 |
+
"past_value_25_in",
|
| 256 |
+
"past_key_26_in",
|
| 257 |
+
"past_value_26_in",
|
| 258 |
+
"past_key_27_in",
|
| 259 |
+
"past_value_27_in",
|
| 260 |
+
"past_key_28_in",
|
| 261 |
+
"past_value_28_in",
|
| 262 |
+
"position_ids_cos",
|
| 263 |
+
"position_ids_sin",
|
| 264 |
+
"attention_mask"
|
| 265 |
+
],
|
| 266 |
+
"outputs": [
|
| 267 |
+
"past_value_24_out",
|
| 268 |
+
"past_key_24_out",
|
| 269 |
+
"past_value_25_out",
|
| 270 |
+
"past_key_25_out",
|
| 271 |
+
"past_value_26_out",
|
| 272 |
+
"past_key_26_out",
|
| 273 |
+
"past_value_27_out",
|
| 274 |
+
"past_key_27_out",
|
| 275 |
+
"past_value_28_out",
|
| 276 |
+
"past_key_28_out",
|
| 277 |
+
"past_value_29_out",
|
| 278 |
+
"past_key_29_out",
|
| 279 |
+
"past_value_30_out",
|
| 280 |
+
"past_key_30_out",
|
| 281 |
+
"past_value_31_out",
|
| 282 |
+
"past_key_31_out",
|
| 283 |
+
"logits"
|
| 284 |
+
],
|
| 285 |
+
"session_options": {
|
| 286 |
+
"log_id": "onnxruntime-genai.pp4",
|
| 287 |
+
"provider_options": [
|
| 288 |
+
{
|
| 289 |
+
"qnn": {
|
| 290 |
+
"backend_path": "QnnHtp.dll",
|
| 291 |
+
"htp_performance_mode": "burst",
|
| 292 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 293 |
+
"qnn_context_priority": "high"
|
| 294 |
+
}
|
| 295 |
+
}
|
| 296 |
+
]
|
| 297 |
+
},
|
| 298 |
+
"run_on_token_gen": false
|
| 299 |
+
},
|
| 300 |
+
"token-generator-1": {
|
| 301 |
+
"filename": "ar1_cl4096_1_of_4_qnn_ctx.onnx",
|
| 302 |
+
"inputs": [
|
| 303 |
+
"input_ids",
|
| 304 |
+
"past_key_0_in",
|
| 305 |
+
"past_key_5_in",
|
| 306 |
+
"past_value_5_in",
|
| 307 |
+
"past_value_0_in",
|
| 308 |
+
"past_key_6_in",
|
| 309 |
+
"past_value_6_in",
|
| 310 |
+
"past_key_7_in",
|
| 311 |
+
"past_value_7_in",
|
| 312 |
+
"past_key_1_in",
|
| 313 |
+
"past_value_1_in",
|
| 314 |
+
"past_key_2_in",
|
| 315 |
+
"past_value_2_in",
|
| 316 |
+
"past_key_3_in",
|
| 317 |
+
"past_value_3_in",
|
| 318 |
+
"past_key_4_in",
|
| 319 |
+
"past_value_4_in",
|
| 320 |
+
"position_ids_cos",
|
| 321 |
+
"position_ids_sin",
|
| 322 |
+
"attention_mask"
|
| 323 |
+
],
|
| 324 |
+
"outputs": [
|
| 325 |
+
"past_value_0_out",
|
| 326 |
+
"past_key_0_out",
|
| 327 |
+
"past_value_1_out",
|
| 328 |
+
"past_key_1_out",
|
| 329 |
+
"past_value_2_out",
|
| 330 |
+
"past_key_2_out",
|
| 331 |
+
"past_value_3_out",
|
| 332 |
+
"past_key_3_out",
|
| 333 |
+
"past_value_4_out",
|
| 334 |
+
"past_key_4_out",
|
| 335 |
+
"past_value_5_out",
|
| 336 |
+
"past_key_5_out",
|
| 337 |
+
"past_value_6_out",
|
| 338 |
+
"past_key_6_out",
|
| 339 |
+
"past_value_7_out",
|
| 340 |
+
"past_key_7_out",
|
| 341 |
+
"_model_layers_7_Add_1_Add_output_0"
|
| 342 |
+
],
|
| 343 |
+
"session_options": {
|
| 344 |
+
"log_id": "onnxruntime-genai.tg1",
|
| 345 |
+
"provider_options": [
|
| 346 |
+
{
|
| 347 |
+
"qnn": {
|
| 348 |
+
"backend_path": "QnnHtp.dll",
|
| 349 |
+
"htp_performance_mode": "burst",
|
| 350 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 351 |
+
"qnn_context_priority": "high"
|
| 352 |
+
}
|
| 353 |
+
}
|
| 354 |
+
]
|
| 355 |
+
},
|
| 356 |
+
"run_on_prompt": false
|
| 357 |
+
},
|
| 358 |
+
"token-generator-2": {
|
| 359 |
+
"filename": "ar1_cl4096_2_of_4_qnn_ctx.onnx",
|
| 360 |
+
"inputs": [
|
| 361 |
+
"_model_layers_7_Add_1_Add_output_0",
|
| 362 |
+
"past_key_8_in",
|
| 363 |
+
"past_key_13_in",
|
| 364 |
+
"past_value_13_in",
|
| 365 |
+
"past_value_8_in",
|
| 366 |
+
"past_key_14_in",
|
| 367 |
+
"past_value_14_in",
|
| 368 |
+
"past_key_15_in",
|
| 369 |
+
"past_value_15_in",
|
| 370 |
+
"past_key_9_in",
|
| 371 |
+
"past_value_9_in",
|
| 372 |
+
"past_key_10_in",
|
| 373 |
+
"past_value_10_in",
|
| 374 |
+
"past_key_11_in",
|
| 375 |
+
"past_value_11_in",
|
| 376 |
+
"past_key_12_in",
|
| 377 |
+
"past_value_12_in",
|
| 378 |
+
"position_ids_cos",
|
| 379 |
+
"position_ids_sin",
|
| 380 |
+
"attention_mask"
|
| 381 |
+
],
|
| 382 |
+
"outputs": [
|
| 383 |
+
"past_value_8_out",
|
| 384 |
+
"past_key_8_out",
|
| 385 |
+
"past_value_9_out",
|
| 386 |
+
"past_key_9_out",
|
| 387 |
+
"past_value_10_out",
|
| 388 |
+
"past_key_10_out",
|
| 389 |
+
"past_value_11_out",
|
| 390 |
+
"past_key_11_out",
|
| 391 |
+
"past_value_12_out",
|
| 392 |
+
"past_key_12_out",
|
| 393 |
+
"past_value_13_out",
|
| 394 |
+
"past_key_13_out",
|
| 395 |
+
"past_value_14_out",
|
| 396 |
+
"past_key_14_out",
|
| 397 |
+
"past_value_15_out",
|
| 398 |
+
"past_key_15_out",
|
| 399 |
+
"_model_layers_15_Add_1_Add_output_0"
|
| 400 |
+
],
|
| 401 |
+
"session_options": {
|
| 402 |
+
"log_id": "onnxruntime-genai.tg2",
|
| 403 |
+
"provider_options": [
|
| 404 |
+
{
|
| 405 |
+
"qnn": {
|
| 406 |
+
"backend_path": "QnnHtp.dll",
|
| 407 |
+
"htp_performance_mode": "burst",
|
| 408 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 409 |
+
"qnn_context_priority": "high"
|
| 410 |
+
}
|
| 411 |
+
}
|
| 412 |
+
]
|
| 413 |
+
},
|
| 414 |
+
"run_on_prompt": false
|
| 415 |
+
},
|
| 416 |
+
"token-generator-3": {
|
| 417 |
+
"filename": "ar1_cl4096_3_of_4_qnn_ctx.onnx",
|
| 418 |
+
"inputs": [
|
| 419 |
+
"_model_layers_15_Add_1_Add_output_0",
|
| 420 |
+
"past_key_16_in",
|
| 421 |
+
"past_key_21_in",
|
| 422 |
+
"past_value_21_in",
|
| 423 |
+
"past_value_16_in",
|
| 424 |
+
"past_key_22_in",
|
| 425 |
+
"past_value_22_in",
|
| 426 |
+
"past_key_23_in",
|
| 427 |
+
"past_value_23_in",
|
| 428 |
+
"past_key_17_in",
|
| 429 |
+
"past_value_17_in",
|
| 430 |
+
"past_key_18_in",
|
| 431 |
+
"past_value_18_in",
|
| 432 |
+
"past_key_19_in",
|
| 433 |
+
"past_value_19_in",
|
| 434 |
+
"past_key_20_in",
|
| 435 |
+
"past_value_20_in",
|
| 436 |
+
"position_ids_cos",
|
| 437 |
+
"position_ids_sin",
|
| 438 |
+
"attention_mask"
|
| 439 |
+
],
|
| 440 |
+
"outputs": [
|
| 441 |
+
"past_value_16_out",
|
| 442 |
+
"past_key_16_out",
|
| 443 |
+
"past_value_17_out",
|
| 444 |
+
"past_key_17_out",
|
| 445 |
+
"past_value_18_out",
|
| 446 |
+
"past_key_18_out",
|
| 447 |
+
"past_value_19_out",
|
| 448 |
+
"past_key_19_out",
|
| 449 |
+
"past_value_20_out",
|
| 450 |
+
"past_key_20_out",
|
| 451 |
+
"past_value_21_out",
|
| 452 |
+
"past_key_21_out",
|
| 453 |
+
"past_value_22_out",
|
| 454 |
+
"past_key_22_out",
|
| 455 |
+
"past_value_23_out",
|
| 456 |
+
"past_key_23_out",
|
| 457 |
+
"_model_layers_23_Add_1_Add_output_0"
|
| 458 |
+
],
|
| 459 |
+
"session_options": {
|
| 460 |
+
"log_id": "onnxruntime-genai.tg3",
|
| 461 |
+
"provider_options": [
|
| 462 |
+
{
|
| 463 |
+
"qnn": {
|
| 464 |
+
"backend_path": "QnnHtp.dll",
|
| 465 |
+
"htp_performance_mode": "burst",
|
| 466 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 467 |
+
"qnn_context_priority": "high"
|
| 468 |
+
}
|
| 469 |
+
}
|
| 470 |
+
]
|
| 471 |
+
},
|
| 472 |
+
"run_on_prompt": false
|
| 473 |
+
},
|
| 474 |
+
"token-generator-4": {
|
| 475 |
+
"filename": "ar1_cl4096_4_of_4_qnn_ctx.onnx",
|
| 476 |
+
"inputs": [
|
| 477 |
+
"_model_layers_23_Add_1_Add_output_0",
|
| 478 |
+
"past_key_24_in",
|
| 479 |
+
"past_key_29_in",
|
| 480 |
+
"past_value_29_in",
|
| 481 |
+
"past_value_24_in",
|
| 482 |
+
"past_key_30_in",
|
| 483 |
+
"past_value_30_in",
|
| 484 |
+
"past_key_31_in",
|
| 485 |
+
"past_value_31_in",
|
| 486 |
+
"past_key_25_in",
|
| 487 |
+
"past_value_25_in",
|
| 488 |
+
"past_key_26_in",
|
| 489 |
+
"past_value_26_in",
|
| 490 |
+
"past_key_27_in",
|
| 491 |
+
"past_value_27_in",
|
| 492 |
+
"past_key_28_in",
|
| 493 |
+
"past_value_28_in",
|
| 494 |
+
"position_ids_cos",
|
| 495 |
+
"position_ids_sin",
|
| 496 |
+
"attention_mask"
|
| 497 |
+
],
|
| 498 |
+
"outputs": [
|
| 499 |
+
"past_value_24_out",
|
| 500 |
+
"past_key_24_out",
|
| 501 |
+
"past_value_25_out",
|
| 502 |
+
"past_key_25_out",
|
| 503 |
+
"past_value_26_out",
|
| 504 |
+
"past_key_26_out",
|
| 505 |
+
"past_value_27_out",
|
| 506 |
+
"past_key_27_out",
|
| 507 |
+
"past_value_28_out",
|
| 508 |
+
"past_key_28_out",
|
| 509 |
+
"past_value_29_out",
|
| 510 |
+
"past_key_29_out",
|
| 511 |
+
"past_value_30_out",
|
| 512 |
+
"past_key_30_out",
|
| 513 |
+
"past_value_31_out",
|
| 514 |
+
"past_key_31_out",
|
| 515 |
+
"logits"
|
| 516 |
+
],
|
| 517 |
+
"session_options": {
|
| 518 |
+
"log_id": "onnxruntime-genai.tg4",
|
| 519 |
+
"provider_options": [
|
| 520 |
+
{
|
| 521 |
+
"qnn": {
|
| 522 |
+
"backend_path": "QnnHtp.dll",
|
| 523 |
+
"htp_performance_mode": "burst",
|
| 524 |
+
"enable_htp_shared_memory_allocator": "1",
|
| 525 |
+
"qnn_context_priority": "high"
|
| 526 |
+
}
|
| 527 |
+
}
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"run_on_prompt": false
|
| 531 |
+
},
|
| 532 |
+
"dequantizer": {
|
| 533 |
+
"filename": "dequantizer.onnx",
|
| 534 |
+
"inputs": [
|
| 535 |
+
"logits"
|
| 536 |
+
],
|
| 537 |
+
"outputs": [
|
| 538 |
+
"logits_dequantized"
|
| 539 |
+
],
|
| 540 |
+
"session_options": {
|
| 541 |
+
"log_id": "onnxruntime-genai.dequantizer",
|
| 542 |
+
"provider_options": [
|
| 543 |
+
{}
|
| 544 |
+
]
|
| 545 |
+
}
|
| 546 |
+
}
|
| 547 |
+
}
|
| 548 |
+
]
|
| 549 |
+
},
|
| 550 |
+
"eos_token_id": [
|
| 551 |
+
32007,
|
| 552 |
+
32001,
|
| 553 |
+
32000,
|
| 554 |
+
2
|
| 555 |
+
],
|
| 556 |
+
"pad_token_id": 32000,
|
| 557 |
+
"type": "decoder-pipeline",
|
| 558 |
+
"vocab_size": 32064
|
| 559 |
+
},
|
| 560 |
+
"search": {
|
| 561 |
+
"diversity_penalty": 0.0,
|
| 562 |
+
"do_sample": true,
|
| 563 |
+
"early_stopping": true,
|
| 564 |
+
"length_penalty": 1.0,
|
| 565 |
+
"max_length": 2048,
|
| 566 |
+
"min_length": 0,
|
| 567 |
+
"no_repeat_ngram_size": 0,
|
| 568 |
+
"num_beams": 1,
|
| 569 |
+
"num_return_sequences": 1,
|
| 570 |
+
"past_present_share_buffer": true,
|
| 571 |
+
"repetition_penalty": 1.0,
|
| 572 |
+
"temperature": 0.6,
|
| 573 |
+
"top_k": 1,
|
| 574 |
+
"top_p": 1.0
|
| 575 |
+
}
|
| 576 |
+
}
|
position-processor.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c339bfd62cdc452be4e5f31efe9681fe7e9635163bb7ca8a56080fdad651202f
|
| 3 |
+
size 15327
|
quantizer.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4af353f75fd7f683304f66fc2e95e89ba12faed2ced42e2ce7a5ba8602b546bf
|
| 3 |
+
size 1147
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|endoftext|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "<|endoftext|>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": true,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": false
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "<|endoftext|>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
},
|
| 38 |
+
"32001": {
|
| 39 |
+
"content": "<|assistant|>",
|
| 40 |
+
"lstrip": false,
|
| 41 |
+
"normalized": false,
|
| 42 |
+
"rstrip": true,
|
| 43 |
+
"single_word": false,
|
| 44 |
+
"special": true
|
| 45 |
+
},
|
| 46 |
+
"32002": {
|
| 47 |
+
"content": "<|placeholder1|>",
|
| 48 |
+
"lstrip": false,
|
| 49 |
+
"normalized": false,
|
| 50 |
+
"rstrip": true,
|
| 51 |
+
"single_word": false,
|
| 52 |
+
"special": true
|
| 53 |
+
},
|
| 54 |
+
"32003": {
|
| 55 |
+
"content": "<|placeholder2|>",
|
| 56 |
+
"lstrip": false,
|
| 57 |
+
"normalized": false,
|
| 58 |
+
"rstrip": true,
|
| 59 |
+
"single_word": false,
|
| 60 |
+
"special": true
|
| 61 |
+
},
|
| 62 |
+
"32004": {
|
| 63 |
+
"content": "<|placeholder3|>",
|
| 64 |
+
"lstrip": false,
|
| 65 |
+
"normalized": false,
|
| 66 |
+
"rstrip": true,
|
| 67 |
+
"single_word": false,
|
| 68 |
+
"special": true
|
| 69 |
+
},
|
| 70 |
+
"32005": {
|
| 71 |
+
"content": "<|placeholder4|>",
|
| 72 |
+
"lstrip": false,
|
| 73 |
+
"normalized": false,
|
| 74 |
+
"rstrip": true,
|
| 75 |
+
"single_word": false,
|
| 76 |
+
"special": true
|
| 77 |
+
},
|
| 78 |
+
"32006": {
|
| 79 |
+
"content": "<|system|>",
|
| 80 |
+
"lstrip": false,
|
| 81 |
+
"normalized": false,
|
| 82 |
+
"rstrip": true,
|
| 83 |
+
"single_word": false,
|
| 84 |
+
"special": true
|
| 85 |
+
},
|
| 86 |
+
"32007": {
|
| 87 |
+
"content": "<|end|>",
|
| 88 |
+
"lstrip": false,
|
| 89 |
+
"normalized": false,
|
| 90 |
+
"rstrip": true,
|
| 91 |
+
"single_word": false,
|
| 92 |
+
"special": true
|
| 93 |
+
},
|
| 94 |
+
"32008": {
|
| 95 |
+
"content": "<|placeholder5|>",
|
| 96 |
+
"lstrip": false,
|
| 97 |
+
"normalized": false,
|
| 98 |
+
"rstrip": true,
|
| 99 |
+
"single_word": false,
|
| 100 |
+
"special": true
|
| 101 |
+
},
|
| 102 |
+
"32009": {
|
| 103 |
+
"content": "<|placeholder6|>",
|
| 104 |
+
"lstrip": false,
|
| 105 |
+
"normalized": false,
|
| 106 |
+
"rstrip": true,
|
| 107 |
+
"single_word": false,
|
| 108 |
+
"special": true
|
| 109 |
+
},
|
| 110 |
+
"32010": {
|
| 111 |
+
"content": "<|user|>",
|
| 112 |
+
"lstrip": false,
|
| 113 |
+
"normalized": false,
|
| 114 |
+
"rstrip": true,
|
| 115 |
+
"single_word": false,
|
| 116 |
+
"special": true
|
| 117 |
+
}
|
| 118 |
+
},
|
| 119 |
+
"bos_token": "<s>",
|
| 120 |
+
"chat_template": "{% for message in messages %}{% if message['role'] == 'system' and message['content'] %}{{'<|system|>\n' + message['content'] + '<|end|>\n'}}{% elif message['role'] == 'user' %}{{'<|user|>\n' + message['content'] + '<|end|>\n'}}{% elif message['role'] == 'assistant' %}{{'<|assistant|>\n' + message['content'] + '<|end|>\n'}}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|assistant|>\n' }}{% else %}{{ eos_token }}{% endif %}",
|
| 121 |
+
"clean_up_tokenization_spaces": false,
|
| 122 |
+
"eos_token": "<|endoftext|>",
|
| 123 |
+
"legacy": false,
|
| 124 |
+
"model_max_length": 131072,
|
| 125 |
+
"pad_token": "<|endoftext|>",
|
| 126 |
+
"padding_side": "left",
|
| 127 |
+
"sp_model_kwargs": {},
|
| 128 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 129 |
+
"unk_token": "<unk>",
|
| 130 |
+
"use_default_system_prompt": false
|
| 131 |
+
}
|
weight_sharing_model_1_of_4.serialized.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
weight_sharing_model_2_of_4.serialized.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
weight_sharing_model_3_of_4.serialized.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
weight_sharing_model_4_of_4.serialized.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|