Improve model card: Add `library_name`, update `pipeline_tag`, and correct `language` entry
Browse filesThis PR enhances the model card by:
- Adding `library_name: transformers` to the metadata, which enables the interactive "how to use" widget on the Hugging Face Hub, making it easier for users to get started with the model. This is supported by the `transformers` library usage in the "Quickstart" section.
- Updating the `pipeline_tag` from `translation` to `text-generation` as per guidelines for generative language models, and adding `translation` as an additional `tag` for better categorization.
- Correcting the invalid `language: false` entry to `language: no` (Norwegian), ensuring the language list accurately reflects the 60 languages supported by the model as detailed in the "Support Languages" section.
README.md
CHANGED
|
@@ -1,4 +1,6 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
language:
|
| 3 |
- en
|
| 4 |
- zh
|
|
@@ -60,10 +62,11 @@ language:
|
|
| 60 |
- ur
|
| 61 |
- uz
|
| 62 |
- yue
|
| 63 |
-
base_model:
|
| 64 |
-
- Qwen/Qwen3-8B-Base
|
| 65 |
license: apache-2.0
|
| 66 |
-
pipeline_tag:
|
|
|
|
|
|
|
|
|
|
| 67 |
---
|
| 68 |
|
| 69 |
## LMT
|
|
@@ -73,7 +76,7 @@ pipeline_tag: translation
|
|
| 73 |
**LMT-60** is a suite of **Chinese-English-centric** MMT models trained on **90B tokens** mixed monolingual and bilingual tokens, covering **60 languages across 234 translation directions** and achieving **SOTA performance** among models with similar language coverage.
|
| 74 |
We release both the CPT and SFT versions of LMT-60 in four sizes (0.6B/1.7B/4B/8B). All checkpoints are available:
|
| 75 |
| Models | Model Link |
|
| 76 |
-
|
| 77 |
| LMT-60-0.6B-Base | [NiuTrans/LMT-60-0.6B-Base](https://huggingface.co/NiuTrans/LMT-60-0.6B-Base) |
|
| 78 |
| LMT-60-0.6B | [NiuTrans/LMT-60-0.6B](https://huggingface.co/NiuTrans/LMT-60-0.6B) |
|
| 79 |
| LMT-60-1.7B-Base | [NiuTrans/LMT-60-1.7B-Base](https://huggingface.co/NiuTrans/LMT-60-1.7B-Base) |
|
|
@@ -95,7 +98,9 @@ model_name = "NiuTrans/LMT-60-8B"
|
|
| 95 |
tokenizer = AutoTokenizer.from_pretrained(model_name, padding_side='left')
|
| 96 |
model = AutoModelForCausalLM.from_pretrained(model_name)
|
| 97 |
|
| 98 |
-
prompt = "Translate the following text from English into Chinese
|
|
|
|
|
|
|
| 99 |
messages = [{"role": "user", "content": prompt}]
|
| 100 |
text = tokenizer.apply_chat_template(
|
| 101 |
messages,
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen3-8B-Base
|
| 4 |
language:
|
| 5 |
- en
|
| 6 |
- zh
|
|
|
|
| 62 |
- ur
|
| 63 |
- uz
|
| 64 |
- yue
|
|
|
|
|
|
|
| 65 |
license: apache-2.0
|
| 66 |
+
pipeline_tag: text-generation
|
| 67 |
+
library_name: transformers
|
| 68 |
+
tags:
|
| 69 |
+
- translation
|
| 70 |
---
|
| 71 |
|
| 72 |
## LMT
|
|
|
|
| 76 |
**LMT-60** is a suite of **Chinese-English-centric** MMT models trained on **90B tokens** mixed monolingual and bilingual tokens, covering **60 languages across 234 translation directions** and achieving **SOTA performance** among models with similar language coverage.
|
| 77 |
We release both the CPT and SFT versions of LMT-60 in four sizes (0.6B/1.7B/4B/8B). All checkpoints are available:
|
| 78 |
| Models | Model Link |
|
| 79 |
+
|:------------|:------------|\
|
| 80 |
| LMT-60-0.6B-Base | [NiuTrans/LMT-60-0.6B-Base](https://huggingface.co/NiuTrans/LMT-60-0.6B-Base) |
|
| 81 |
| LMT-60-0.6B | [NiuTrans/LMT-60-0.6B](https://huggingface.co/NiuTrans/LMT-60-0.6B) |
|
| 82 |
| LMT-60-1.7B-Base | [NiuTrans/LMT-60-1.7B-Base](https://huggingface.co/NiuTrans/LMT-60-1.7B-Base) |
|
|
|
|
| 98 |
tokenizer = AutoTokenizer.from_pretrained(model_name, padding_side='left')
|
| 99 |
model = AutoModelForCausalLM.from_pretrained(model_name)
|
| 100 |
|
| 101 |
+
prompt = "Translate the following text from English into Chinese.
|
| 102 |
+
English: The concept came from China where plum blossoms were the flower of choice.
|
| 103 |
+
Chinese: "
|
| 104 |
messages = [{"role": "user", "content": prompt}]
|
| 105 |
text = tokenizer.apply_chat_template(
|
| 106 |
messages,
|