Instructions to use hutlim/Qwen3-Reranker-0.6B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use hutlim/Qwen3-Reranker-0.6B with Transformers:
# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("hutlim/Qwen3-Reranker-0.6B") model = AutoModelForCausalLM.from_pretrained("hutlim/Qwen3-Reranker-0.6B") - Notebooks
- Google Colab
- Kaggle
Update handler.py
Browse files- handler.py +2 -2
handler.py
CHANGED
|
@@ -39,7 +39,7 @@ class EndpointHandler:
|
|
| 39 |
)
|
| 40 |
|
| 41 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 42 |
-
self.
|
| 43 |
|
| 44 |
self.tokenizer = AutoTokenizer.from_pretrained(
|
| 45 |
str(model_dir),
|
|
@@ -49,7 +49,7 @@ class EndpointHandler:
|
|
| 49 |
|
| 50 |
self.model = AutoModelForCausalLM.from_pretrained(
|
| 51 |
str(model_dir),
|
| 52 |
-
|
| 53 |
trust_remote_code=True,
|
| 54 |
).to(self.device).eval()
|
| 55 |
|
|
|
|
| 39 |
)
|
| 40 |
|
| 41 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 42 |
+
self.torch_dtype = torch.float16 if self.device == "cuda" else torch.float32
|
| 43 |
|
| 44 |
self.tokenizer = AutoTokenizer.from_pretrained(
|
| 45 |
str(model_dir),
|
|
|
|
| 49 |
|
| 50 |
self.model = AutoModelForCausalLM.from_pretrained(
|
| 51 |
str(model_dir),
|
| 52 |
+
torch_dtype=self.torch_dtype,
|
| 53 |
trust_remote_code=True,
|
| 54 |
).to(self.device).eval()
|
| 55 |
|