Update README.md
Browse files
README.md
CHANGED
@@ -15,13 +15,13 @@ metrics:
|
|
15 |
- comet
|
16 |
pipeline_tag: translation
|
17 |
---
|
18 |
-
# Model Card for TowerInstruct-7B-v0.
|
19 |
|
20 |
## Model Details
|
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.
|
25 |
The model is trained to handle several translation-related tasks, such as general machine translation (e.g., sentence- and paragraph/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 |
|
@@ -31,7 +31,7 @@ We will release more details in the upcoming technical report.
|
|
31 |
- **License:** CC-BY-NC-4.0, Llama 2 is licensed under the [LLAMA 2 Community License](https://ai.meta.com/llama/license/), Copyright © Meta Platforms, Inc. All Rights Reserved.
|
32 |
- **Finetuned from model:** [TowerBase](https://huggingface.co/Unbabel/TowerBase-7B-v0.1)
|
33 |
|
34 |
-
**Update**: TowerInstruct-7B-v0.2 has more reliable document-level translation capabilities in comparison with TowerInstruct-7B-v0.
|
35 |
|
36 |
## Intended uses & limitations
|
37 |
|
@@ -59,7 +59,7 @@ Here's how you can run the model using the `pipeline()` function from 🤗 Trans
|
|
59 |
import torch
|
60 |
from transformers import pipeline
|
61 |
|
62 |
-
pipe = pipeline("text-generation", model="Unbabel/TowerInstruct-v0.
|
63 |
# We use the tokenizer’s chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
|
64 |
messages = [
|
65 |
{"role": "user", "content": "Translate the following text from Portuguese into English.\nPortuguese: Um grupo de investigadores lançou um novo modelo para tarefas relacionadas com tradução.\nEnglish:"},
|
@@ -82,11 +82,11 @@ We are currently working on improving quality and consistency on document-level
|
|
82 |
|
83 |
## Bias, Risks, and Limitations
|
84 |
|
85 |
-
TowerInstruct-v0.
|
86 |
|
87 |
## Prompt Format
|
88 |
|
89 |
-
TowerInstruct-v0.
|
90 |
```
|
91 |
<|im_start|>user
|
92 |
{USER PROMPT}<|im_end|>
|
|
|
15 |
- comet
|
16 |
pipeline_tag: translation
|
17 |
---
|
18 |
+
# Model Card for TowerInstruct-7B-v0.2
|
19 |
|
20 |
## Model Details
|
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-v0.2 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 paragraph/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 |
|
|
|
31 |
- **License:** CC-BY-NC-4.0, Llama 2 is licensed under the [LLAMA 2 Community License](https://ai.meta.com/llama/license/), Copyright © Meta Platforms, Inc. All Rights Reserved.
|
32 |
- **Finetuned from model:** [TowerBase](https://huggingface.co/Unbabel/TowerBase-7B-v0.1)
|
33 |
|
34 |
+
**Update**: TowerInstruct-7B-v0.2 has more reliable document-level translation capabilities in comparison with TowerInstruct-7B-v0.2. The new version of TowerBlocks used to train v0.2 is also available in the Tower collection.
|
35 |
|
36 |
## Intended uses & limitations
|
37 |
|
|
|
59 |
import torch
|
60 |
from transformers import pipeline
|
61 |
|
62 |
+
pipe = pipeline("text-generation", model="Unbabel/TowerInstruct-v0.2", torch_dtype=torch.bfloat16, device_map="auto")
|
63 |
# We use the tokenizer’s chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
|
64 |
messages = [
|
65 |
{"role": "user", "content": "Translate the following text from Portuguese into English.\nPortuguese: Um grupo de investigadores lançou um novo modelo para tarefas relacionadas com tradução.\nEnglish:"},
|
|
|
82 |
|
83 |
## Bias, Risks, and Limitations
|
84 |
|
85 |
+
TowerInstruct-v0.2 has not been aligned to human preferences, so the model may generate problematic outputs (e.g., hallucinations, harmful content, or false statements).
|
86 |
|
87 |
## Prompt Format
|
88 |
|
89 |
+
TowerInstruct-v0.2 was trained using the ChatML prompt templates without any system prompts. An example follows below:
|
90 |
```
|
91 |
<|im_start|>user
|
92 |
{USER PROMPT}<|im_end|>
|