Update README.md
Browse files
README.md
CHANGED
|
@@ -21,7 +21,7 @@ pipeline_tag: translation
|
|
| 21 |
|
| 22 |
### Model Description
|
| 23 |
|
| 24 |
-
TowerInstruct is a language model that results from fine-tuning TowerBase on the TowerBlocks supervised fine-tuning dataset. TowerInstruct
|
| 25 |
The model is trained to handle several translation-related tasks, such as general machine translation (e.g., sentence- and document-level translation, terminology-aware translation, context-aware translation), automatic post edition, named-entity recognition, gramatical error correction, and paraphrase generation.
|
| 26 |
We will release more details in the upcoming technical report.
|
| 27 |
|
|
@@ -29,7 +29,7 @@ We will release more details in the upcoming technical report.
|
|
| 29 |
- **Model type:** A 7B parameter model fine-tuned on a mix of publicly available, synthetic datasets on translation-related tasks, as well as conversational datasets and code instructions.
|
| 30 |
- **Language(s) (NLP):** English, Portuguese, Spanish, French, German, Dutch, Italian, Korean, Chinese, Russian
|
| 31 |
- **License:** CC-BY-NC-4.0
|
| 32 |
-
- **Finetuned from model:** TowerBase
|
| 33 |
|
| 34 |
## Intended uses & limitations
|
| 35 |
|
|
@@ -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 here.
|
| 49 |
|
| 50 |
Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
|
| 51 |
|
|
@@ -95,37 +95,15 @@ TowerInstruct-v0.1 was trained using the ChatML prompt templates without any sys
|
|
| 95 |
|
| 96 |
### Supervised tasks
|
| 97 |
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
Translate the following text from $SRC_LANG into $TGT_LANG.
|
| 101 |
-
$SRC_LANG: $SRC_TEXT
|
| 102 |
-
$TGT_LANG: # make sure to add a white space the target placeholder "$TGT_LANG:" for best results
|
| 103 |
-
```
|
| 104 |
-
- Automatic Post Edition
|
| 105 |
-
```
|
| 106 |
-
Translate the following text from $SRC_LANG into $TGT_LANG.
|
| 107 |
-
$SRC_LANG: $SRC_TEXT
|
| 108 |
-
$TGT_LANG:
|
| 109 |
-
```
|
| 110 |
-
- Machine Translation Evaluation
|
| 111 |
-
- Context-aware Translation
|
| 112 |
-
- Terminology-aware Translation
|
| 113 |
-
- Multi-reference Translation
|
| 114 |
-
- Named-entity Recognition
|
| 115 |
-
- Paraphrase Generation
|
| 116 |
-
- Synthetic Chat data
|
| 117 |
-
- Code instructions
|
| 118 |
[More Information Needed]
|
| 119 |
|
| 120 |
## Training Details
|
| 121 |
|
| 122 |
### Training Data
|
| 123 |
|
| 124 |
-
Link to TowerBlocks.
|
| 125 |
-
|
| 126 |
-
### Training Procedure
|
| 127 |
-
|
| 128 |
-
Write sth about Axolotl.
|
| 129 |
|
| 130 |
#### Training Hyperparameters
|
| 131 |
|
|
|
|
| 21 |
|
| 22 |
### Model Description
|
| 23 |
|
| 24 |
+
TowerInstruct-7B is a language model that results from fine-tuning TowerBase on the TowerBlocks supervised fine-tuning dataset. TowerInstruct-7B-v0.1 is the first model in the series.
|
| 25 |
The model is trained to handle several translation-related tasks, such as general machine translation (e.g., sentence- and document-level translation, terminology-aware translation, context-aware translation), automatic post edition, named-entity recognition, gramatical error correction, and paraphrase generation.
|
| 26 |
We will release more details in the upcoming technical report.
|
| 27 |
|
|
|
|
| 29 |
- **Model type:** A 7B parameter model fine-tuned on a mix of publicly available, synthetic datasets on translation-related tasks, as well as conversational datasets and code instructions.
|
| 30 |
- **Language(s) (NLP):** English, Portuguese, Spanish, French, German, Dutch, Italian, Korean, Chinese, Russian
|
| 31 |
- **License:** CC-BY-NC-4.0
|
| 32 |
+
- **Finetuned from model:** TowerBase [ADD LINK]
|
| 33 |
|
| 34 |
## Intended uses & limitations
|
| 35 |
|
|
|
|
| 45 |
- Synthetic Chat data
|
| 46 |
- Code instructions
|
| 47 |
|
| 48 |
+
You can find the dataset and all data sources of TowerBlocks [ADD LINK] here.
|
| 49 |
|
| 50 |
Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
|
| 51 |
|
|
|
|
| 95 |
|
| 96 |
### Supervised tasks
|
| 97 |
|
| 98 |
+
The prompts for all supervised tasks can be found in TowerBlocks [ADD LINK]. 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.
|
| 99 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
[More Information Needed]
|
| 101 |
|
| 102 |
## Training Details
|
| 103 |
|
| 104 |
### Training Data
|
| 105 |
|
| 106 |
+
Link to TowerBlocks [ADD LINK].
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
|
| 108 |
#### Training Hyperparameters
|
| 109 |
|