Upload 3 files
Browse files- MiniLMv6.cs +61 -36
- MiniLMv6.sentis +2 -2
- info.json +2 -2
MiniLMv6.cs
CHANGED
@@ -3,6 +3,7 @@ using UnityEngine;
|
|
3 |
using Unity.Sentis;
|
4 |
using System.IO;
|
5 |
using System.Text;
|
|
|
6 |
|
7 |
/*
|
8 |
* Tiny Stories Inference Code
|
@@ -35,40 +36,43 @@ public class MiniLM : MonoBehaviour
|
|
35 |
const int START_TOKEN = 101;
|
36 |
const int END_TOKEN = 102;
|
37 |
|
38 |
-
Ops ops;
|
39 |
-
ITensorAllocator allocator;
|
40 |
-
|
41 |
//Store the vocabulary
|
42 |
string[] tokens;
|
43 |
|
44 |
-
|
|
|
|
|
45 |
|
46 |
void Start()
|
47 |
{
|
48 |
-
allocator = new TensorCachingAllocator();
|
49 |
-
ops = WorkerFactory.CreateOps(backend, allocator);
|
50 |
-
|
51 |
tokens = File.ReadAllLines(Application.streamingAssetsPath + "/vocab.txt");
|
52 |
|
53 |
-
|
54 |
|
55 |
-
|
56 |
|
57 |
var tokens1 = GetTokens(string1);
|
58 |
var tokens2 = GetTokens(string2);
|
59 |
|
60 |
-
TensorFloat embedding1 = GetEmbedding(tokens1);
|
61 |
-
TensorFloat embedding2 = GetEmbedding(tokens2);
|
|
|
|
|
62 |
|
63 |
-
Debug.Log("Similarity Score: " +
|
64 |
}
|
65 |
|
66 |
-
float
|
67 |
{
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
|
|
|
|
|
|
|
|
|
|
72 |
}
|
73 |
|
74 |
TensorFloat GetEmbedding(List<int> tokens)
|
@@ -85,31 +89,52 @@ public class MiniLM : MonoBehaviour
|
|
85 |
|
86 |
var inputs = new Dictionary<string, Tensor>
|
87 |
{
|
88 |
-
{"
|
89 |
-
{"
|
90 |
-
{"
|
91 |
};
|
92 |
|
93 |
engine.Execute(inputs);
|
94 |
|
95 |
-
var
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
96 |
|
97 |
-
return
|
98 |
}
|
99 |
|
100 |
//Get average of token embeddings taking into account the attention mask
|
101 |
-
|
102 |
{
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
}
|
114 |
|
115 |
List<int> GetTokens(string text)
|
@@ -150,9 +175,9 @@ public class MiniLM : MonoBehaviour
|
|
150 |
}
|
151 |
|
152 |
private void OnDestroy()
|
153 |
-
{
|
|
|
154 |
engine?.Dispose();
|
155 |
-
ops?.Dispose();
|
156 |
-
allocator?.Dispose();
|
157 |
}
|
|
|
158 |
}
|
|
|
3 |
using Unity.Sentis;
|
4 |
using System.IO;
|
5 |
using System.Text;
|
6 |
+
using FF = Unity.Sentis.Functional;
|
7 |
|
8 |
/*
|
9 |
* Tiny Stories Inference Code
|
|
|
36 |
const int START_TOKEN = 101;
|
37 |
const int END_TOKEN = 102;
|
38 |
|
|
|
|
|
|
|
39 |
//Store the vocabulary
|
40 |
string[] tokens;
|
41 |
|
42 |
+
const int FEATURES = 384; //size of feature space
|
43 |
+
|
44 |
+
IWorker engine, dotScore;
|
45 |
|
46 |
void Start()
|
47 |
{
|
|
|
|
|
|
|
48 |
tokens = File.ReadAllLines(Application.streamingAssetsPath + "/vocab.txt");
|
49 |
|
50 |
+
engine = CreateMLModel();
|
51 |
|
52 |
+
dotScore = CreateDotScoreModel();
|
53 |
|
54 |
var tokens1 = GetTokens(string1);
|
55 |
var tokens2 = GetTokens(string2);
|
56 |
|
57 |
+
using TensorFloat embedding1 = GetEmbedding(tokens1);
|
58 |
+
using TensorFloat embedding2 = GetEmbedding(tokens2);
|
59 |
+
|
60 |
+
float score = GetDotScore(embedding1, embedding2);
|
61 |
|
62 |
+
Debug.Log("Similarity Score: " + score);
|
63 |
}
|
64 |
|
65 |
+
float GetDotScore(TensorFloat A, TensorFloat B)
|
66 |
{
|
67 |
+
var inputs = new Dictionary<string, Tensor>()
|
68 |
+
{
|
69 |
+
{ "input_0", A },
|
70 |
+
{ "input_1", B }
|
71 |
+
};
|
72 |
+
dotScore.Execute(inputs);
|
73 |
+
var output = dotScore.PeekOutput() as TensorFloat;
|
74 |
+
output.CompleteOperationsAndDownload();
|
75 |
+
return output[0];
|
76 |
}
|
77 |
|
78 |
TensorFloat GetEmbedding(List<int> tokens)
|
|
|
89 |
|
90 |
var inputs = new Dictionary<string, Tensor>
|
91 |
{
|
92 |
+
{"input_0", input_ids },
|
93 |
+
{"input_1", attention_mask },
|
94 |
+
{"input_2", token_type_ids}
|
95 |
};
|
96 |
|
97 |
engine.Execute(inputs);
|
98 |
|
99 |
+
var output = engine.TakeOutputOwnership("output_0") as TensorFloat;
|
100 |
+
return output;
|
101 |
+
}
|
102 |
+
|
103 |
+
IWorker CreateMLModel()
|
104 |
+
{
|
105 |
+
Model model = ModelLoader.Load(Application.streamingAssetsPath + "/MiniLMv6.sentis");
|
106 |
+
|
107 |
+
Model modelWithMeanPooling = Functional.Compile(
|
108 |
+
(input_ids, attention_mask, token_type_ids) =>
|
109 |
+
{
|
110 |
+
var tokenEmbeddings = model.Forward(input_ids, attention_mask, token_type_ids)[0];
|
111 |
+
return MeanPooling(tokenEmbeddings, attention_mask);
|
112 |
+
},
|
113 |
+
(model.inputs[0], model.inputs[1], model.inputs[2])
|
114 |
+
);
|
115 |
|
116 |
+
return WorkerFactory.CreateWorker(backend, modelWithMeanPooling);
|
117 |
}
|
118 |
|
119 |
//Get average of token embeddings taking into account the attention mask
|
120 |
+
FunctionalTensor MeanPooling(FunctionalTensor tokenEmbeddings, FunctionalTensor attentionMask)
|
121 |
{
|
122 |
+
var mask = attentionMask.Unsqueeze(-1).BroadcastTo(new[] { FEATURES }); //shape=(1,N,FEATURES)
|
123 |
+
var A = FF.ReduceSum(tokenEmbeddings * mask, 1, false); //shape=(1,FEATURES)
|
124 |
+
var B = A / (FF.ReduceSum(mask, 1, false) + 1e-9f); //shape=(1,FEATURES)
|
125 |
+
var C = FF.Sqrt(FF.ReduceSum(FF.Square(B), 1, true)); //shape=(1,FEATURES)
|
126 |
+
return B / C; //shape=(1,FEATURES)
|
127 |
+
}
|
128 |
+
|
129 |
+
IWorker CreateDotScoreModel()
|
130 |
+
{
|
131 |
+
Model dotScoreModel = Functional.Compile(
|
132 |
+
(input1, input2) => Functional.ReduceSum(input1 * input2, 1),
|
133 |
+
(InputDef.Float(new TensorShape(1, FEATURES)),
|
134 |
+
InputDef.Float(new TensorShape(1, FEATURES)))
|
135 |
+
);
|
136 |
+
|
137 |
+
return WorkerFactory.CreateWorker(backend, dotScoreModel);
|
138 |
}
|
139 |
|
140 |
List<int> GetTokens(string text)
|
|
|
175 |
}
|
176 |
|
177 |
private void OnDestroy()
|
178 |
+
{
|
179 |
+
dotScore?.Dispose();
|
180 |
engine?.Dispose();
|
|
|
|
|
181 |
}
|
182 |
+
|
183 |
}
|
MiniLMv6.sentis
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c9a2597ce9edce4c09b32e993b7f906cce91fceb2f461a597b974f71ee70453d
|
3 |
+
size 90898400
|
info.json
CHANGED
@@ -8,7 +8,7 @@
|
|
8 |
"data": [
|
9 |
"vocab.txt"
|
10 |
],
|
11 |
-
"version":[
|
12 |
-
"1.
|
13 |
]
|
14 |
}
|
|
|
8 |
"data": [
|
9 |
"vocab.txt"
|
10 |
],
|
11 |
+
"version": [
|
12 |
+
"1.4.0"
|
13 |
]
|
14 |
}
|