Diwank Singh
commited on
Commit
•
5f36a54
1
Parent(s):
fcca1db
v2
Browse filesSigned-off-by: Diwank Singh <diwank.singh@gmail.com>
- 2_Asym/140225464223632_Dense/config.json +1 -0
- 2_Asym/140225464223632_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464231312_Dense/config.json +1 -0
- 2_Asym/140225464231312_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464232416_Dense/config.json +1 -0
- 2_Asym/140225464232416_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464233232_Dropout/config.json +1 -0
- 2_Asym/140225464235392_Dense/config.json +1 -0
- 2_Asym/140225464235392_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464235968_Dense/config.json +1 -0
- 2_Asym/140225464235968_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464236064_Dense/config.json +1 -0
- 2_Asym/140225464236064_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225464236784_Dense/config.json +1 -0
- 2_Asym/140225464236784_Dense/pytorch_model.bin +3 -0
- 2_Asym/140225507851568_Dropout/config.json +1 -0
- 2_Asym/140228101622896_Dense/config.json +1 -0
- 2_Asym/140228101622896_Dense/pytorch_model.bin +3 -0
- 2_Asym/config.json +24 -16
- README.md +14 -6
2_Asym/140225464223632_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 1024, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464223632_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e09d27f6e37e79b5889f99320b29eab442407414661f86706af8946f783d0239
|
3 |
+
size 8398460
|
2_Asym/140225464231312_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464231312_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:75db6e76d6b5838f1cd255ba48a1fffb13b807985583df180effbfb73abbfc92
|
3 |
+
size 16787068
|
2_Asym/140225464232416_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 1024, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464232416_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f73d9d5ee30e1c9c367620f4b32b59133841a9aa66c34c213c4eceeed1d91739
|
3 |
+
size 8394364
|
2_Asym/140225464233232_Dropout/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"dropout": 0.1}
|
2_Asym/140225464235392_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464235392_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:53306f081b130f877d2d20e684b403951a08cf410c6854f2fc23f49427d5ba85
|
3 |
+
size 16787068
|
2_Asym/140225464235968_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 1024, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464235968_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:762e434b5b7dcfef9ae42841a4193578ecb74d750c5efa5433fd0b129883511f
|
3 |
+
size 8394364
|
2_Asym/140225464236064_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464236064_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b7cc7b864e5a7d41a179fa8660606d0d65e7a0ae3a3ac2076b810ee07234861c
|
3 |
+
size 16787068
|
2_Asym/140225464236784_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 1024, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140225464236784_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f2759d8139196720ab46d120d3225231d160bd9fa74f39d7f59bab8a447155db
|
3 |
+
size 8398460
|
2_Asym/140225507851568_Dropout/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"dropout": 0.1}
|
2_Asym/140228101622896_Dense/config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"in_features": 2048, "out_features": 2048, "bias": true, "activation_function": "torch.nn.modules.activation.Tanh"}
|
2_Asym/140228101622896_Dense/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5c82ba0624007b1e5b7f281fdafa92e7106f7c7f954de0617fcf3ff261bd154f
|
3 |
+
size 16787068
|
2_Asym/config.json
CHANGED
@@ -1,26 +1,34 @@
|
|
1 |
{
|
2 |
"types": {
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
8 |
-
"
|
9 |
-
"
|
10 |
-
"
|
|
|
|
|
|
|
|
|
11 |
},
|
12 |
"structure": {
|
13 |
"dialog": [
|
14 |
-
"
|
15 |
-
"
|
16 |
-
"
|
17 |
-
"
|
|
|
|
|
18 |
],
|
19 |
"fact": [
|
20 |
-
"
|
21 |
-
"
|
22 |
-
"
|
23 |
-
"
|
|
|
|
|
24 |
]
|
25 |
},
|
26 |
"parameters": {
|
|
|
1 |
{
|
2 |
"types": {
|
3 |
+
"140225464236784_Dense": "sentence_transformers.models.Dense",
|
4 |
+
"140228101622896_Dense": "sentence_transformers.models.Dense",
|
5 |
+
"140225507851568_Dropout": "sentence_transformers.models.Dropout",
|
6 |
+
"140225464235392_Dense": "sentence_transformers.models.Dense",
|
7 |
+
"140225464235968_Dense": "sentence_transformers.models.Dense",
|
8 |
+
"140225464231168_Normalize": "sentence_transformers.models.Normalize",
|
9 |
+
"140225464223632_Dense": "sentence_transformers.models.Dense",
|
10 |
+
"140225464236064_Dense": "sentence_transformers.models.Dense",
|
11 |
+
"140225464233232_Dropout": "sentence_transformers.models.Dropout",
|
12 |
+
"140225464231312_Dense": "sentence_transformers.models.Dense",
|
13 |
+
"140225464232416_Dense": "sentence_transformers.models.Dense",
|
14 |
+
"140225464235536_Normalize": "sentence_transformers.models.Normalize"
|
15 |
},
|
16 |
"structure": {
|
17 |
"dialog": [
|
18 |
+
"140225464236784_Dense",
|
19 |
+
"140228101622896_Dense",
|
20 |
+
"140225507851568_Dropout",
|
21 |
+
"140225464235392_Dense",
|
22 |
+
"140225464235968_Dense",
|
23 |
+
"140225464231168_Normalize"
|
24 |
],
|
25 |
"fact": [
|
26 |
+
"140225464223632_Dense",
|
27 |
+
"140225464236064_Dense",
|
28 |
+
"140225464233232_Dropout",
|
29 |
+
"140225464231312_Dense",
|
30 |
+
"140225464232416_Dense",
|
31 |
+
"140225464235536_Normalize"
|
32 |
]
|
33 |
},
|
34 |
"parameters": {
|
README.md
CHANGED
@@ -53,12 +53,12 @@ The model was trained with the parameters:
|
|
53 |
|
54 |
**Loss**:
|
55 |
|
56 |
-
`sentence_transformers.losses.
|
57 |
|
58 |
Parameters of the fit()-Method:
|
59 |
```
|
60 |
{
|
61 |
-
"epochs":
|
62 |
"evaluation_steps": 2000,
|
63 |
"evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
|
64 |
"max_grad_norm": 1,
|
@@ -83,12 +83,20 @@ SentenceTransformer(
|
|
83 |
(2): Asym(
|
84 |
(dialog-0): Dense({'in_features': 1024, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
85 |
(dialog-1): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
86 |
-
(dialog-2):
|
87 |
-
|
|
|
|
|
|
|
|
|
88 |
(fact-0): Dense({'in_features': 1024, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
89 |
(fact-1): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
90 |
-
(fact-2):
|
91 |
-
|
|
|
|
|
|
|
|
|
92 |
)
|
93 |
)
|
94 |
```
|
|
|
53 |
|
54 |
**Loss**:
|
55 |
|
56 |
+
`sentence_transformers.losses.CosineSimilarityLoss.CosineSimilarityLoss`
|
57 |
|
58 |
Parameters of the fit()-Method:
|
59 |
```
|
60 |
{
|
61 |
+
"epochs": 4,
|
62 |
"evaluation_steps": 2000,
|
63 |
"evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
|
64 |
"max_grad_norm": 1,
|
|
|
83 |
(2): Asym(
|
84 |
(dialog-0): Dense({'in_features': 1024, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
85 |
(dialog-1): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
86 |
+
(dialog-2): Dropout(
|
87 |
+
(dropout_layer): Dropout(p=0.1, inplace=False)
|
88 |
+
)
|
89 |
+
(dialog-3): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
90 |
+
(dialog-4): Dense({'in_features': 2048, 'out_features': 1024, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
91 |
+
(dialog-5): Normalize()
|
92 |
(fact-0): Dense({'in_features': 1024, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
93 |
(fact-1): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
94 |
+
(fact-2): Dropout(
|
95 |
+
(dropout_layer): Dropout(p=0.1, inplace=False)
|
96 |
+
)
|
97 |
+
(fact-3): Dense({'in_features': 2048, 'out_features': 2048, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
98 |
+
(fact-4): Dense({'in_features': 2048, 'out_features': 1024, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
99 |
+
(fact-5): Normalize()
|
100 |
)
|
101 |
)
|
102 |
```
|