Instructions to use TigreGotico/opus-mt-en-az-onnx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use TigreGotico/opus-mt-en-az-onnx with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "translation" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("translation", model="TigreGotico/opus-mt-en-az-onnx")# Load model directly from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("TigreGotico/opus-mt-en-az-onnx") model = AutoModelForSeq2SeqLM.from_pretrained("TigreGotico/opus-mt-en-az-onnx", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -50,6 +50,14 @@ print(tok.decode(out[0], skip_special_tokens=True))
|
|
| 50 |
10 general-domain sentences, exact-string-match of generated output against
|
| 51 |
`MarianMTModel.generate()` on the original `Helsinki-NLP/opus-mt-en-az` checkpoint.
|
| 52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
| Decoding | fp32 exact match | int8 exact match |
|
| 54 |
|---|---|---|
|
| 55 |
| greedy (num_beams=1) | 10/10 (100.0%) | 9/10 (90.0%) |
|
|
|
|
| 50 |
10 general-domain sentences, exact-string-match of generated output against
|
| 51 |
`MarianMTModel.generate()` on the original `Helsinki-NLP/opus-mt-en-az` checkpoint.
|
| 52 |
|
| 53 |
+
```yaml
|
| 54 |
+
parity:
|
| 55 |
+
fp32_greedy: 1.00 # 10/10
|
| 56 |
+
fp32_beam4: 1.00 # 10/10
|
| 57 |
+
int8_greedy: 0.90 # 9/10
|
| 58 |
+
int8_beam4: 0.40 # 4/10
|
| 59 |
+
```
|
| 60 |
+
|
| 61 |
| Decoding | fp32 exact match | int8 exact match |
|
| 62 |
|---|---|---|
|
| 63 |
| greedy (num_beams=1) | 10/10 (100.0%) | 9/10 (90.0%) |
|