Update README.md
Browse files
README.md
CHANGED
|
@@ -33,7 +33,7 @@ We will release more details in the upcoming technical report. For now, you can
|
|
| 33 |
|
| 34 |
## Intended uses & limitations
|
| 35 |
|
| 36 |
-
The model was initially fine-tuned on a filtered and preprocessed supervised fine-tuning dataset ([TowerBlocks](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.
|
| 37 |
- Translation (sentence and paragraph-level)
|
| 38 |
- Automatic Post Edition
|
| 39 |
- Machine Translation Evaluation
|
|
@@ -45,7 +45,7 @@ The model was initially fine-tuned on a filtered and preprocessed supervised fin
|
|
| 45 |
- Synthetic Chat data
|
| 46 |
- Code instructions
|
| 47 |
|
| 48 |
-
You can find the dataset and all data sources of [TowerBlocks](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.
|
| 49 |
|
| 50 |
Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
|
| 51 |
|
|
@@ -96,13 +96,13 @@ TowerInstruct-v0.1 was trained using the ChatML prompt templates without any sys
|
|
| 96 |
|
| 97 |
### Supervised tasks
|
| 98 |
|
| 99 |
-
The prompts for all supervised tasks can be found in [TowerBlocks](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.
|
| 100 |
|
| 101 |
## Training Details
|
| 102 |
|
| 103 |
### Training Data
|
| 104 |
|
| 105 |
-
Link to [TowerBlocks](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.
|
| 106 |
|
| 107 |
#### Training Hyperparameters
|
| 108 |
|
|
|
|
| 33 |
|
| 34 |
## Intended uses & limitations
|
| 35 |
|
| 36 |
+
The model was initially fine-tuned on a filtered and preprocessed supervised fine-tuning dataset ([TowerBlocks-v0.2](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.2)), which contains a diverse range of data sources:
|
| 37 |
- Translation (sentence and paragraph-level)
|
| 38 |
- Automatic Post Edition
|
| 39 |
- Machine Translation Evaluation
|
|
|
|
| 45 |
- Synthetic Chat data
|
| 46 |
- Code instructions
|
| 47 |
|
| 48 |
+
You can find the dataset and all data sources of [TowerBlocks-v0.2](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.2) here.
|
| 49 |
|
| 50 |
Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
|
| 51 |
|
|
|
|
| 96 |
|
| 97 |
### Supervised tasks
|
| 98 |
|
| 99 |
+
The prompts for all supervised tasks can be found in [TowerBlocks-v0.2](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.2). We have used multiple prompt templates for each task. While different prompts may offer different outputs, the difference in downstream performance should be very minimal.
|
| 100 |
|
| 101 |
## Training Details
|
| 102 |
|
| 103 |
### Training Data
|
| 104 |
|
| 105 |
+
Link to [TowerBlocks-v0.2](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.2).
|
| 106 |
|
| 107 |
#### Training Hyperparameters
|
| 108 |
|