Hugging Face's logo

IlyaGusev
/
mbart_ru_sum_gazeta

Summarization
Transformers
PyTorch
Safetensors
Russian
mbart
text2text-generation
Inference Endpoints
Model card Files Community
2
IlyaGusev commited on
Commit
5df2e50
•
1 Parent(s): 599d3c6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -12
README.md CHANGED
@@ -51,7 +51,7 @@ input_ids = tokenizer(
51
 
52
  output_ids = model.generate(
53
  input_ids=input_ids,
54
- no_repeat_ngram_size=3
55
  )[0]
56
 
57
  summary = tokenizer.decode(output_ids, skip_special_tokens=True)
@@ -78,26 +78,26 @@ print(summary)
78
  * Test dataset: **Gazeta v1 test**
79
  * Source max_length: **600**
80
  * Target max_length: **200**
81
- * no_repeat_ngram_size: **3**
82
  * num_beams: **5**
83
 
84
- | Model | R-1-f | R-2-f | R-L-f | chrF | METEOR | BLEU |
85
- |:--------------------------|:------|:------|:------|:-------|:-------|:-----|
86
- | [mbart_ru_sum_gazeta](https://huggingface.co/IlyaGusev/mbart_ru_sum_gazeta) | **32.3** | 14.3 | **27.9** | **39.8** | **26.3** | **12.2** |
87
- | [rut5_base_sum_gazeta](https://huggingface.co/IlyaGusev/rut5_base_sum_gazeta) | 32.2 | **14.4** | 27.8 | 39.6 | 25.3 | 11.9 |
88
-
89
 
90
  * Train dataset: **Gazeta v1 train**
91
  * Test dataset: **Gazeta v2 test**
92
  * Source max_length: **600**
93
  * Target max_length: **200**
94
- * no_repeat_ngram_size: **3**
95
  * num_beams: **5**
96
 
97
- | Model | R-1-f | R-2-f | R-L-f | chrF | METEOR | BLEU |
98
- |:--------------------------|:------|:------|:------|:-------|:-------|:-----|
99
- | [mbart_ru_sum_gazeta](https://huggingface.co/IlyaGusev/mbart_ru_sum_gazeta) | **28.8** | **11.1** | **24.5** | **37.4** | **22.7** | **9.5** |
100
- | [rut5_base_sum_gazeta](https://huggingface.co/IlyaGusev/rut5_base_sum_gazeta) | 28.5 | 11.0 | 24.2 | 36.8 | 21.2 | 8.6 |
101
 
102
  Predicting all summaries:
103
  ```python
 
51
 
52
  output_ids = model.generate(
53
  input_ids=input_ids,
54
+ no_repeat_ngram_size=4
55
  )[0]
56
 
57
  summary = tokenizer.decode(output_ids, skip_special_tokens=True)
 
78
  * Test dataset: **Gazeta v1 test**
79
  * Source max_length: **600**
80
  * Target max_length: **200**
81
+ * no_repeat_ngram_size: **4**
82
  * num_beams: **5**
83
 
84
+ | Model | R-1-f | R-2-f | R-L-f | chrF | METEOR | BLEU | Avg char length |
85
+ |:--------------------------|:------|:------|:------|:-------|:-------|:-----|:-----|
86
+ | [mbart_ru_sum_gazeta](https://huggingface.co/IlyaGusev/mbart_ru_sum_gazeta) | **32.4** | 14.3 | 28.0 | 39.7 | **26.4** | 12.1 | 371 |
87
+ | [rut5_base_sum_gazeta](https://huggingface.co/IlyaGusev/rut5_base_sum_gazeta) | 32.2 | **14.4** | **28.1** | **39.8** | 25.7 | **12.3** | 330 |
88
+ |
89
 
90
  * Train dataset: **Gazeta v1 train**
91
  * Test dataset: **Gazeta v2 test**
92
  * Source max_length: **600**
93
  * Target max_length: **200**
94
+ * no_repeat_ngram_size: **4**
95
  * num_beams: **5**
96
 
97
+ | Model | R-1-f | R-2-f | R-L-f | chrF | METEOR | BLEU | Avg char length |
98
+ |:--------------------------|:------|:------|:------|:-------|:-------|:-----|:-----|
99
+ | [mbart_ru_sum_gazeta](https://huggingface.co/IlyaGusev/mbart_ru_sum_gazeta) | **28.7** | **11.1** | **24.4** | **37.3** | **22.7** | **9.4** | 373 |
100
+ | [rut5_base_sum_gazeta](https://huggingface.co/IlyaGusev/rut5_base_sum_gazeta) | 28.6 | 11.1 | **24.5** | 37.2 | 22.0 | 9.4 | 331 |
101
 
102
  Predicting all summaries:
103
  ```python