Spaces:
Running
Running
Minor
Browse files
README.md
CHANGED
@@ -27,6 +27,7 @@ computes precision, recall and F1 scores.
|
|
27 |
## How to Use
|
28 |
|
29 |
Sem-F1 takes 2 mandatory arguments:
|
|
|
30 |
- `predictions` - List of predictions. Format varies based on `tokenize_sentences` and `multi_references` flags.
|
31 |
- `references`: List of references. Format varies based on `tokenize_sentences` and `multi_references` flags.
|
32 |
|
@@ -49,19 +50,20 @@ for score in results:
|
|
49 |
|
50 |
|
51 |
Sem-F1 also accepts multiple optional arguments:
|
52 |
-
- `model_type (str)`: Model to use for encoding sentences. Options: ['pv1', 'stsb', 'use']
|
53 |
-
- `pv1` - [paraphrase-distilroberta-base-v1](https://huggingface.co/sentence-transformers/paraphrase-distilroberta-base-v1)
|
54 |
-
- `stsb` - [stsb-roberta-large](https://huggingface.co/sentence-transformers/stsb-roberta-large)
|
55 |
-
- `use` - [Universal Sentence Encoder](https://huggingface.co/sentence-transformers/use-cmlm-multilingual) (Default)
|
56 |
|
57 |
-
|
58 |
-
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
|
66 |
Refer to the inputs descriptions for more detailed usage as follows:
|
67 |
|
@@ -78,6 +80,7 @@ print(metric.inputs_description)
|
|
78 |
|
79 |
### Output Values
|
80 |
List of `Scores` dataclass corresponding to each sample -
|
|
|
81 |
- `precision: float`: Precision score, which ranges from 0.0 to 1.0.
|
82 |
- `recall: List[float]`: Recall score corresponding to each reference
|
83 |
- `f1: float`: F1 score (between precision and average recall).
|
|
|
27 |
## How to Use
|
28 |
|
29 |
Sem-F1 takes 2 mandatory arguments:
|
30 |
+
|
31 |
- `predictions` - List of predictions. Format varies based on `tokenize_sentences` and `multi_references` flags.
|
32 |
- `references`: List of references. Format varies based on `tokenize_sentences` and `multi_references` flags.
|
33 |
|
|
|
50 |
|
51 |
|
52 |
Sem-F1 also accepts multiple optional arguments:
|
|
|
|
|
|
|
|
|
53 |
|
54 |
+
- `model_type (str)`: Model to use for encoding sentences. Options: ['pv1', 'stsb', 'use']
|
55 |
+
- `pv1` - [paraphrase-distilroberta-base-v1](https://huggingface.co/sentence-transformers/paraphrase-distilroberta-base-v1)
|
56 |
+
- `stsb` - [stsb-roberta-large](https://huggingface.co/sentence-transformers/stsb-roberta-large)
|
57 |
+
- `use` - [Universal Sentence Encoder](https://huggingface.co/sentence-transformers/use-cmlm-multilingual) (Default)
|
58 |
+
|
59 |
+
Furthermore, you can use any model on Huggingface/SentenceTransformer that is supported by SentenceTransformer
|
60 |
+
such as `all-mpnet-base-v2` or `roberta-base`
|
61 |
|
62 |
+
- `tokenize_sentences (bool)`: Flag to indicate whether to tokenize the sentences in the input documents. Default: True.
|
63 |
+
- `multi_references (bool)`: Flag to indicate whether multiple references are provided. Default: False.
|
64 |
+
- `gpu (Union[bool, str, int, List[Union[str, int]]])`: Whether to use GPU, CPU or multiple-processes for computation.
|
65 |
+
- `batch_size (int)`: Batch size for encoding. Default: 32.
|
66 |
+
- `verbose (bool)`: Flag to indicate verbose output. Default: False.
|
67 |
|
68 |
Refer to the inputs descriptions for more detailed usage as follows:
|
69 |
|
|
|
80 |
|
81 |
### Output Values
|
82 |
List of `Scores` dataclass corresponding to each sample -
|
83 |
+
|
84 |
- `precision: float`: Precision score, which ranges from 0.0 to 1.0.
|
85 |
- `recall: List[float]`: Recall score corresponding to each reference
|
86 |
- `f1: float`: F1 score (between precision and average recall).
|