hmm404/triplet_728
Browse files
README.md
CHANGED
@@ -4,60 +4,52 @@ tags:
|
|
4 |
- sentence-similarity
|
5 |
- feature-extraction
|
6 |
- generated_from_trainer
|
7 |
-
- dataset_size:
|
8 |
- loss:TripletLoss
|
9 |
base_model: sentence-transformers/all-mpnet-base-v2
|
10 |
widget:
|
11 |
-
- source_sentence:
|
|
|
12 |
sentences:
|
13 |
-
-
|
14 |
-
|
15 |
-
-
|
16 |
-
|
17 |
-
-
|
18 |
-
|
19 |
-
- source_sentence:
|
20 |
-
|
|
|
21 |
sentences:
|
22 |
-
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
cost and hazard -- in effect penalize society rather than the murderer who committed
|
32 |
-
the crime.
|
33 |
-
- 2 Nephi 3:6-22 in the Book of Mormon describes Joseph Smith as a "choice seer."
|
34 |
-
- source_sentence: Daenerys has proven that she is incapable of ruling even a single
|
35 |
-
city.
|
36 |
sentences:
|
37 |
-
-
|
38 |
-
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
- source_sentence: Consumers within public health systems should be empowered to make
|
43 |
-
their own choices about their treatment.
|
44 |
sentences:
|
45 |
-
-
|
46 |
-
|
47 |
-
-
|
48 |
-
|
49 |
-
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
sentences:
|
54 |
-
-
|
55 |
-
|
56 |
-
|
57 |
-
-
|
58 |
-
the realm of the "is."
|
59 |
-
- Putting a child up for adoption can be considered a selfless act if you do it
|
60 |
-
because someone else can give the child a better life than you can.
|
61 |
pipeline_tag: sentence-similarity
|
62 |
library_name: sentence-transformers
|
63 |
---
|
@@ -112,9 +104,9 @@ from sentence_transformers import SentenceTransformer
|
|
112 |
model = SentenceTransformer("sentence_transformers_model_id")
|
113 |
# Run inference
|
114 |
sentences = [
|
115 |
-
|
116 |
-
'
|
117 |
-
'
|
118 |
]
|
119 |
embeddings = model.encode(sentences)
|
120 |
print(embeddings.shape)
|
@@ -168,24 +160,24 @@ You can finetune this model on your own dataset.
|
|
168 |
|
169 |
#### Unnamed Dataset
|
170 |
|
171 |
-
* Size:
|
172 |
* Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
|
173 |
* Approximate statistics based on the first 1000 samples:
|
174 |
| | anchor | positive | negative |
|
175 |
|:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
|
176 |
| type | string | string | string |
|
177 |
-
| details | <ul><li>min:
|
178 |
* Samples:
|
179 |
-
| anchor
|
180 |
-
|
181 |
-
| <code>
|
182 |
-
| <code>
|
183 |
-
| <code>
|
184 |
* Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
|
185 |
```json
|
186 |
{
|
187 |
"distance_metric": "TripletDistanceMetric.COSINE",
|
188 |
-
"triplet_margin": 0.
|
189 |
}
|
190 |
```
|
191 |
|
@@ -313,27 +305,29 @@ You can finetune this model on your own dataset.
|
|
313 |
</details>
|
314 |
|
315 |
### Training Logs
|
316 |
-
| Epoch | Step
|
317 |
-
|
318 |
-
| 0.
|
319 |
-
| 0.
|
320 |
-
| 0.
|
321 |
-
| 0.
|
322 |
-
| 0.
|
323 |
-
| 0.
|
324 |
-
|
|
325 |
-
| 1.
|
326 |
-
| 1.
|
327 |
-
| 1.
|
328 |
-
| 1.
|
329 |
-
| 1.
|
330 |
-
|
|
331 |
-
|
|
332 |
-
| 2.
|
333 |
-
| 2.
|
334 |
-
| 2.
|
335 |
-
| 2.
|
336 |
-
| 2.
|
|
|
|
|
337 |
|
338 |
|
339 |
### Framework Versions
|
|
|
4 |
- sentence-similarity
|
5 |
- feature-extraction
|
6 |
- generated_from_trainer
|
7 |
+
- dataset_size:29115
|
8 |
- loss:TripletLoss
|
9 |
base_model: sentence-transformers/all-mpnet-base-v2
|
10 |
widget:
|
11 |
+
- source_sentence: China doesn't employ enough African executives or workers to spread
|
12 |
+
know-how
|
13 |
sentences:
|
14 |
+
- Africans will have better knowledge of infrastructure when they have it in their
|
15 |
+
backyards, even if they aren't the ones who work on it.
|
16 |
+
- The best way of learning is to be taught. African companies need to be involved
|
17 |
+
a lot more into the process of developing infrastructure.
|
18 |
+
- Just 36 percent say they would vote to reelect the president in 2020, compared
|
19 |
+
with 44 percent who would pick the Democrat.
|
20 |
+
- source_sentence: If determinism is true, then the our understanding that determinism
|
21 |
+
is true is based on chance conditions of physics and chemistry, not reason and
|
22 |
+
logic. If determinism is false, the conclusion is obviously incorrect.
|
23 |
sentences:
|
24 |
+
- Predetermination does not invalidate the proven effectiveness of logic. In fact
|
25 |
+
causal determinism demands it. Thus determinism does not invalidate our conclusions.
|
26 |
+
- Some children are not ready to follow strict schedules at such a young age.
|
27 |
+
- Without free will, the scientist cannot choose what conclusions to draw from these
|
28 |
+
observations, but must conclude what she is determined to conclude. Therefore
|
29 |
+
such conclusions cannot be trusted.
|
30 |
+
- source_sentence: Reason alone cannot determine if God exists, so if one chooses
|
31 |
+
to believe in God, one does so knowing that there is a chance it is not true;
|
32 |
+
everyone who believes in God surely accepts this.
|
|
|
|
|
|
|
|
|
|
|
33 |
sentences:
|
34 |
+
- It is immoral to support immoral deities.
|
35 |
+
- Countries would reap socio-economic benefit from the legalisation of drugs.
|
36 |
+
- Believing that there's an all-Knowing and all-Seeing being makes it likelier for
|
37 |
+
human beings to adhere to a moral code which is beneficial for the planet.
|
38 |
+
- source_sentence: The Book of Mormon contains false prophecies.
|
|
|
|
|
39 |
sentences:
|
40 |
+
- The Book of Mormon falsely prophecies that those who seek to destroy Joseph Smith
|
41 |
+
"shall be confounded." \(2 Nephi 3:14\)
|
42 |
+
- Prophecies are vague and metaphorical. Therefore, claiming they are false does
|
43 |
+
not create a strong argument against authenticity.
|
44 |
+
- The hormonal and psychological shifts that occur during pregnancy put some women
|
45 |
+
at an increased risk for mental illnesses, such as anxiety and depression.
|
46 |
+
- source_sentence: The emergence of brands re-establishes a hierarchy within the equalized
|
47 |
+
consumer society.
|
48 |
sentences:
|
49 |
+
- Not everyone can buy very expensive brands and thus need to remain behind in their
|
50 |
+
conspicuous consumption.
|
51 |
+
- Conspicuous consumption has existed before the advent of a consumer society.
|
52 |
+
- Income inequality is extremely high in America.
|
|
|
|
|
|
|
53 |
pipeline_tag: sentence-similarity
|
54 |
library_name: sentence-transformers
|
55 |
---
|
|
|
104 |
model = SentenceTransformer("sentence_transformers_model_id")
|
105 |
# Run inference
|
106 |
sentences = [
|
107 |
+
'The emergence of brands re-establishes a hierarchy within the equalized consumer society.',
|
108 |
+
'Not everyone can buy very expensive brands and thus need to remain behind in their conspicuous consumption.',
|
109 |
+
'Conspicuous consumption has existed before the advent of a consumer society.',
|
110 |
]
|
111 |
embeddings = model.encode(sentences)
|
112 |
print(embeddings.shape)
|
|
|
160 |
|
161 |
#### Unnamed Dataset
|
162 |
|
163 |
+
* Size: 29,115 training samples
|
164 |
* Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
|
165 |
* Approximate statistics based on the first 1000 samples:
|
166 |
| | anchor | positive | negative |
|
167 |
|:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
|
168 |
| type | string | string | string |
|
169 |
+
| details | <ul><li>min: 4 tokens</li><li>mean: 26.7 tokens</li><li>max: 186 tokens</li></ul> | <ul><li>min: 8 tokens</li><li>mean: 34.75 tokens</li><li>max: 176 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 34.92 tokens</li><li>max: 169 tokens</li></ul> |
|
170 |
* Samples:
|
171 |
+
| anchor | positive | negative |
|
172 |
+
|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
|
173 |
+
| <code>There are irrevocable consequences to execution, such as wrongful conviction.</code> | <code>Capital punishment ignores the plausibility that eventually exonerating evidence may come to light. It therefore undermines justice by exacting a punishment that cannot be undone and/or satisfactorily compensated.</code> | <code>Society is willing to accept some limited individual injustices where the system fails in order to protect the overall functioning of justice for most people. As the saying goes: "perfection is the enemy of the good." There are many cases where society accepts some risk of individual deaths in order to protect the public good, such as in the medical system where the risk of accidental death is tolerated, or in traffic safety systems or air travel which tolerate some level of lethal risk, etc.</code> |
|
174 |
+
| <code>Painful executions would be easier and cheaper.</code> | <code>Executions should be administered as effectively as possible, independent of the form they take. If that involves pain, where pain isn't the end goal of the act of punishment, that doesn't seem to be unacceptable.</code> | <code>There is no cost not worthy of taking care for the well-being of citizens, depraved or not, even with respect for capital punishment as a government should be responsible for the well-being of all its citizens.</code> |
|
175 |
+
| <code>Governments are not able to control population growth anyway. Nature instead will take care of it automatically once the carrying capacity of Earth is reached.</code> | <code>It is unrealistic for the government to be expected to consistently provide social goods of high quality in an equitable manner if the population is growing too quickly or shrinking too quickly.</code> | <code>But at what cost? As we approach the "natural carrying capacity" of our planet, we threaten the existence of many animal species.</code> |
|
176 |
* Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
|
177 |
```json
|
178 |
{
|
179 |
"distance_metric": "TripletDistanceMetric.COSINE",
|
180 |
+
"triplet_margin": 0.5
|
181 |
}
|
182 |
```
|
183 |
|
|
|
305 |
</details>
|
306 |
|
307 |
### Training Logs
|
308 |
+
| Epoch | Step | Training Loss |
|
309 |
+
|:------:|:-----:|:-------------:|
|
310 |
+
| 0.1374 | 500 | 0.35 |
|
311 |
+
| 0.2747 | 1000 | 0.3519 |
|
312 |
+
| 0.4121 | 1500 | 0.3335 |
|
313 |
+
| 0.5495 | 2000 | 0.3282 |
|
314 |
+
| 0.6868 | 2500 | 0.3208 |
|
315 |
+
| 0.8242 | 3000 | 0.3168 |
|
316 |
+
| 0.9615 | 3500 | 0.3003 |
|
317 |
+
| 1.0989 | 4000 | 0.263 |
|
318 |
+
| 1.2363 | 4500 | 0.2516 |
|
319 |
+
| 1.3736 | 5000 | 0.2326 |
|
320 |
+
| 1.5110 | 5500 | 0.2405 |
|
321 |
+
| 1.6484 | 6000 | 0.2339 |
|
322 |
+
| 1.7857 | 6500 | 0.2316 |
|
323 |
+
| 1.9231 | 7000 | 0.2302 |
|
324 |
+
| 2.0604 | 7500 | 0.1848 |
|
325 |
+
| 2.1978 | 8000 | 0.149 |
|
326 |
+
| 2.3352 | 8500 | 0.1573 |
|
327 |
+
| 2.4725 | 9000 | 0.1528 |
|
328 |
+
| 2.6099 | 9500 | 0.1617 |
|
329 |
+
| 2.7473 | 10000 | 0.1466 |
|
330 |
+
| 2.8846 | 10500 | 0.1459 |
|
331 |
|
332 |
|
333 |
### Framework Versions
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 437967672
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b97f610cb1366754f40fbb3b2073107185e4061978007e89db2716895aee860e
|
3 |
size 437967672
|
runs/Feb18_13-54-25_dacc3055ce59/events.out.tfevents.1739886870.dacc3055ce59.282.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5118f49bb605689284d63e2a46597cb6121392d88d9c81389d14df48e9a14710
|
3 |
+
size 9160
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5560
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59072cee6c460f5494470eef113e36b67cf5984ee85d45bab5c850defb856740
|
3 |
size 5560
|