Update README.md
Browse files
README.md
CHANGED
|
@@ -46,7 +46,7 @@ Both models share the same input/output specification:
|
|
| 46 |
|---|---|---|---|
|
| 47 |
| Input 1 | `input_ids` | (1, 512) | INT32 |
|
| 48 |
| Input 2 | `attention_mask` | (1, 512) | INT32 |
|
| 49 |
-
| Output | `score` | (1, 1) | FLOAT16 (
|
| 50 |
|
| 51 |
The model expects standard BERT pair tokenization:
|
| 52 |
|
|
|
|
| 46 |
|---|---|---|---|
|
| 47 |
| Input 1 | `input_ids` | (1, 512) | INT32 |
|
| 48 |
| Input 2 | `attention_mask` | (1, 512) | INT32 |
|
| 49 |
+
| Output | `score` | (1, 1) | FLOAT16 (FP16 model) / FLOAT32 (FP32 model) |
|
| 50 |
|
| 51 |
The model expects standard BERT pair tokenization:
|
| 52 |
|