Update README.md
Browse files
README.md
CHANGED
|
@@ -7,13 +7,25 @@ Using the [DistilRoBERTa](https://huggingface.co/distilroberta-base) model as st
|
|
| 7 |
|
| 8 |
### Climate performance model card
|
| 9 |
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
|
| 14 |
-
|
|
| 15 |
-
|
|
| 16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
### BibTeX entry and citation info
|
| 19 |
|
| 7 |
|
| 8 |
### Climate performance model card
|
| 9 |
|
| 10 |
+
|
| 11 |
+
| Minimum card | |
|
| 12 |
+
|--------------------------------------------------------------------------|----------------|
|
| 13 |
+
| 1. Is the resulting model publicly available? | Yes |
|
| 14 |
+
| 2. How much time does the training of the final model take? | 8 hours |
|
| 15 |
+
| 3. How much time did all experiments take (incl. hyperparameter search)? | 288 hours |
|
| 16 |
+
| 4. What was the energy consumption (GPU/CPU)? | 0.7 kW |
|
| 17 |
+
| 5. At which geo location were the computations performed? | Germany |
|
| 18 |
+
|
| 19 |
+
| Extended card | |
|
| 20 |
+
|--------------------------------------------------------------------------|----------------|
|
| 21 |
+
| 6. What was the energy mix at the geo location? | 470 gCO2eq/kWh |
|
| 22 |
+
| 7. How much CO2eq was emitted to train the final model? | 2.63 kg |
|
| 23 |
+
| 8. How much CO2eq was emitted for all experiments? | 94.75 kg |
|
| 24 |
+
| 9. What is the average CO2eq emission for the inference of one sample? | 0.62 mg |
|
| 25 |
+
| 10. Which positive environmental impact can be expected from this work? | This work can be categorized as a building block tools following Jin et al (2021). It supports the training of NLP models in the field of climate change and, thereby, have a positive environmental impact in the future. |
|
| 26 |
+
| 11. Comments | Block pruning could decrease CO2eq emissions |
|
| 27 |
+
|
| 28 |
+
|
| 29 |
|
| 30 |
### BibTeX entry and citation info
|
| 31 |
|