ColeH0415 commited on
Commit
1b5ab9a
·
verified ·
1 Parent(s): cf0a816

CE fine-tuned epoch 2/3

Browse files
README.md CHANGED
@@ -28,25 +28,25 @@ model-index:
28
  type: ce-val
29
  metrics:
30
  - type: accuracy
31
- value: 0.8130841121495327
32
  name: Accuracy
33
  - type: accuracy_threshold
34
- value: -0.2579485774040222
35
  name: Accuracy Threshold
36
  - type: f1
37
- value: 0.8845265588914549
38
  name: F1
39
  - type: f1_threshold
40
- value: -0.43757033348083496
41
  name: F1 Threshold
42
  - type: precision
43
- value: 0.8362445414847162
44
  name: Precision
45
  - type: recall
46
- value: 0.9387254901960784
47
  name: Recall
48
  - type: average_precision
49
- value: 0.9335939639352934
50
  name: Average Precision
51
  ---
52
 
@@ -99,25 +99,25 @@ from sentence_transformers import CrossEncoder
99
  model = CrossEncoder("cross_encoder_model_id")
100
  # Get scores for pairs of inputs
101
  pairs = [
102
- ['Yet a study published just this week, by the Bjerknes Centre for Climate Research in Bergen, Norway, found that the natural climate system can change abruptly, without the need for any external forces.', "Analysis of the layering and chemical composition of the cores has provided a revolutionary new record of climate change in the Northern Hemisphere going back about 100,000\xa0years and illustrated that the world's weather and temperature have often shifted rapidly from one seemingly stable state to another, with worldwide consequences."],
103
- ['Schmittner finds low climate sensitivity.', 'Herbaceous perennials are also able to tolerate the extremes of cold in temperate and Arctic winters, with less sensitivity than trees or shrubs.'],
104
- ['Meeting the 2025 emissions reduction target alone could subtract $250 billion from our GDP and eliminate 2.7 million jobs.', 'Judging by the continued growth in the Renewable Fuel Standard and the extension of the biodiesel tax incentive, the number of jobs can increase to 50,725, $2.7 billion in income, and reaching $5 billion in GDP by 2012 and 2013.'],
105
- ['Preventing global warming is relatively cheap; business-as-usual will cause accelerating climate damage costs that economists struggle to even estimate.', 'Researchers have warned that current economic modeling may seriously underestimate the impact of potentially catastrophic climate change and point to the need for new models that give a more accurate picture of potential damages.'],
106
- ['Domino-effect of climate events could move Earth into a ‘hothouse’ state', 'This one change creates a domino effect which alters many events and characters in the DC Universe.'],
107
  ]
108
  scores = model.predict(pairs)
109
  print(scores)
110
- # [ 3.0222 1.1202 2.4127 1.3557 -0.496 ]
111
 
112
  # Or rank different texts based on similarity to a single text
113
  ranks = model.rank(
114
- 'Yet a study published just this week, by the Bjerknes Centre for Climate Research in Bergen, Norway, found that the natural climate system can change abruptly, without the need for any external forces.',
115
  [
116
- "Analysis of the layering and chemical composition of the cores has provided a revolutionary new record of climate change in the Northern Hemisphere going back about 100,000\xa0years and illustrated that the world's weather and temperature have often shifted rapidly from one seemingly stable state to another, with worldwide consequences.",
117
- 'Herbaceous perennials are also able to tolerate the extremes of cold in temperate and Arctic winters, with less sensitivity than trees or shrubs.',
118
- 'Judging by the continued growth in the Renewable Fuel Standard and the extension of the biodiesel tax incentive, the number of jobs can increase to 50,725, $2.7 billion in income, and reaching $5 billion in GDP by 2012 and 2013.',
119
- 'Researchers have warned that current economic modeling may seriously underestimate the impact of potentially catastrophic climate change and point to the need for new models that give a more accurate picture of potential damages.',
120
- 'This one change creates a domino effect which alters many events and characters in the DC Universe.',
121
  ]
122
  )
123
  # [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]
@@ -158,13 +158,13 @@ You can finetune this model on your own dataset.
158
 
159
  | Metric | Value |
160
  |:----------------------|:-----------|
161
- | accuracy | 0.8131 |
162
- | accuracy_threshold | -0.2579 |
163
- | f1 | 0.8845 |
164
- | f1_threshold | -0.4376 |
165
- | precision | 0.8362 |
166
- | recall | 0.9387 |
167
- | **average_precision** | **0.9336** |
168
 
169
  <!--
170
  ## Bias, Risks and Limitations
@@ -187,16 +187,16 @@ You can finetune this model on your own dataset.
187
  * Size: 4,815 training samples
188
  * Columns: <code>sentence_0</code>, <code>sentence_1</code>, and <code>label</code>
189
  * Approximate statistics based on the first 1000 samples:
190
- | | sentence_0 | sentence_1 | label |
191
- |:--------|:---------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:---------------------------------------------------------------|
192
- | type | string | string | float |
193
- | details | <ul><li>min: 7 tokens</li><li>mean: 26.5 tokens</li><li>max: 66 tokens</li></ul> | <ul><li>min: 7 tokens</li><li>mean: 34.21 tokens</li><li>max: 333 tokens</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.77</li><li>max: 1.0</li></ul> |
194
  * Samples:
195
- | sentence_0 | sentence_1 | label |
196
- |:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------|
197
- | <code>Yet a study published just this week, by the Bjerknes Centre for Climate Research in Bergen, Norway, found that the natural climate system can change abruptly, without the need for any external forces.</code> | <code>Analysis of the layering and chemical composition of the cores has provided a revolutionary new record of climate change in the Northern Hemisphere going back about 100,000 years and illustrated that the world's weather and temperature have often shifted rapidly from one seemingly stable state to another, with worldwide consequences.</code> | <code>1.0</code> |
198
- | <code>Schmittner finds low climate sensitivity.</code> | <code>Herbaceous perennials are also able to tolerate the extremes of cold in temperate and Arctic winters, with less sensitivity than trees or shrubs.</code> | <code>1.0</code> |
199
- | <code>Meeting the 2025 emissions reduction target alone could subtract $250 billion from our GDP and eliminate 2.7 million jobs.</code> | <code>Judging by the continued growth in the Renewable Fuel Standard and the extension of the biodiesel tax incentive, the number of jobs can increase to 50,725, $2.7 billion in income, and reaching $5 billion in GDP by 2012 and 2013.</code> | <code>1.0</code> |
200
  * Loss: [<code>BinaryCrossEntropyLoss</code>](https://sbert.net/docs/package_reference/cross_encoder/losses.html#binarycrossentropyloss) with these parameters:
201
  ```json
202
  {
@@ -317,11 +317,11 @@ You can finetune this model on your own dataset.
317
  ### Training Logs
318
  | Epoch | Step | ce-val_average_precision |
319
  |:-----:|:----:|:------------------------:|
320
- | 1.0 | 301 | 0.9336 |
321
 
322
 
323
  ### Training Time
324
- - **Training**: 24.8 seconds
325
 
326
  ### Framework Versions
327
  - Python: 3.12.13
 
28
  type: ce-val
29
  metrics:
30
  - type: accuracy
31
+ value: 0.8299065420560747
32
  name: Accuracy
33
  - type: accuracy_threshold
34
+ value: -0.3896324336528778
35
  name: Accuracy Threshold
36
  - type: f1
37
+ value: 0.8914285714285715
38
  name: F1
39
  - type: f1_threshold
40
+ value: -1.0047539472579956
41
  name: F1 Threshold
42
  - type: precision
43
+ value: 0.8351177730192719
44
  name: Precision
45
  - type: recall
46
+ value: 0.9558823529411765
47
  name: Recall
48
  - type: average_precision
49
+ value: 0.9535681876950206
50
  name: Average Precision
51
  ---
52
 
 
99
  model = CrossEncoder("cross_encoder_model_id")
100
  # Get scores for pairs of inputs
101
  pairs = [
102
+ ['Climate change is a hoax invented by the Chinese.', '"The concept of global warming was created by and for the Chinese in order to make U.S. manufacturing non-competitive" (Tweet).'],
103
+ ['“The oceans, which absorb more than 90% of the extra CO2 pumped into the atmosphere“', 'Most of the CO 2 taken up by the ocean, which is about 30% of the total released into the atmosphere, forms carbonic acid in equilibrium with bicarbonate.'],
104
+ ['“The jet stream forms a boundary between the cold north and the warmer south, but the lower temperature difference means the winds are now weaker.', 'Therefore, the strong eastward moving jet streams are in part a simple consequence of the fact that the Equator is warmer than the North and South poles.'],
105
+ ['climate models predict too much warming in the troposphere', 'While the satellite data now show global warming, there is still some difference between what climate models predict and what the satellite data show for warming of the lower troposphere, with the climate models predicting slightly more warming than what the satellites measure.'],
106
+ ['Nine years into that 11-year hurricane drought, a NASA scientist computed it as a 1-in-177-year event.', 'It is approximately 177 light years from the Earth.'],
107
  ]
108
  scores = model.predict(pairs)
109
  print(scores)
110
+ # [ 3.6842 2.0192 1.9189 -0.9618 0.3676]
111
 
112
  # Or rank different texts based on similarity to a single text
113
  ranks = model.rank(
114
+ 'Climate change is a hoax invented by the Chinese.',
115
  [
116
+ '"The concept of global warming was created by and for the Chinese in order to make U.S. manufacturing non-competitive" (Tweet).',
117
+ 'Most of the CO 2 taken up by the ocean, which is about 30% of the total released into the atmosphere, forms carbonic acid in equilibrium with bicarbonate.',
118
+ 'Therefore, the strong eastward moving jet streams are in part a simple consequence of the fact that the Equator is warmer than the North and South poles.',
119
+ 'While the satellite data now show global warming, there is still some difference between what climate models predict and what the satellite data show for warming of the lower troposphere, with the climate models predicting slightly more warming than what the satellites measure.',
120
+ 'It is approximately 177 light years from the Earth.',
121
  ]
122
  )
123
  # [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]
 
158
 
159
  | Metric | Value |
160
  |:----------------------|:-----------|
161
+ | accuracy | 0.8299 |
162
+ | accuracy_threshold | -0.3896 |
163
+ | f1 | 0.8914 |
164
+ | f1_threshold | -1.0048 |
165
+ | precision | 0.8351 |
166
+ | recall | 0.9559 |
167
+ | **average_precision** | **0.9536** |
168
 
169
  <!--
170
  ## Bias, Risks and Limitations
 
187
  * Size: 4,815 training samples
188
  * Columns: <code>sentence_0</code>, <code>sentence_1</code>, and <code>label</code>
189
  * Approximate statistics based on the first 1000 samples:
190
+ | | sentence_0 | sentence_1 | label |
191
+ |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:---------------------------------------------------------------|
192
+ | type | string | string | float |
193
+ | details | <ul><li>min: 7 tokens</li><li>mean: 27.21 tokens</li><li>max: 73 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 35.1 tokens</li><li>max: 333 tokens</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.78</li><li>max: 1.0</li></ul> |
194
  * Samples:
195
+ | sentence_0 | sentence_1 | label |
196
+ |:----------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------|
197
+ | <code>Climate change is a hoax invented by the Chinese.</code> | <code>"The concept of global warming was created by and for the Chinese in order to make U.S. manufacturing non-competitive" (Tweet).</code> | <code>1.0</code> |
198
+ | <code>“The oceans, which absorb more than 90% of the extra CO2 pumped into the atmosphere“</code> | <code>Most of the CO 2 taken up by the ocean, which is about 30% of the total released into the atmosphere, forms carbonic acid in equilibrium with bicarbonate.</code> | <code>1.0</code> |
199
+ | <code>“The jet stream forms a boundary between the cold north and the warmer south, but the lower temperature difference means the winds are now weaker.</code> | <code>Therefore, the strong eastward moving jet streams are in part a simple consequence of the fact that the Equator is warmer than the North and South poles.</code> | <code>1.0</code> |
200
  * Loss: [<code>BinaryCrossEntropyLoss</code>](https://sbert.net/docs/package_reference/cross_encoder/losses.html#binarycrossentropyloss) with these parameters:
201
  ```json
202
  {
 
317
  ### Training Logs
318
  | Epoch | Step | ce-val_average_precision |
319
  |:-----:|:----:|:------------------------:|
320
+ | 1.0 | 301 | 0.9536 |
321
 
322
 
323
  ### Training Time
324
+ - **Training**: 27.9 seconds
325
 
326
  ### Framework Versions
327
  - Python: 3.12.13
eval/CrossEncoderClassificationEvaluator_ce-val_results.csv CHANGED
@@ -1,2 +1,3 @@
1
  epoch,steps,Accuracy,Accuracy_Threshold,F1,F1_Threshold,Precision,Recall,Average_Precision
2
  1.0,301,0.8130841121495327,-0.25794858,0.8845265588914549,-0.43757033,0.8362445414847162,0.9387254901960784,0.9335939639352934
 
 
1
  epoch,steps,Accuracy,Accuracy_Threshold,F1,F1_Threshold,Precision,Recall,Average_Precision
2
  1.0,301,0.8130841121495327,-0.25794858,0.8845265588914549,-0.43757033,0.8362445414847162,0.9387254901960784,0.9335939639352934
3
+ 1.0,301,0.8299065420560747,-0.38963243,0.8914285714285715,-1.004754,0.8351177730192719,0.9558823529411765,0.9535681876950206
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:160442534ea4f7d804bce531ea9d64d7765f92e6df6c366ebfb0002c2043e133
3
  size 90866404
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e0ca103c6d4a5cde8fa579d1f6f9007149e9cede51d84a9a59e524ea135df68
3
  size 90866404