ASA
commited on
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
@@ -14,37 +14,5 @@ pipeline_tag: summarization
|
|
14 |
|
15 |
# Meeting-Mind: Meeting Summarization Model
|
16 |
|
17 |
-
This model was fine-tuned on the SAMSum dataset to generate concise summaries of meeting transcripts.
|
18 |
It uses a BART base model with 2 epochs of training.
|
19 |
-
|
20 |
-
## Model Description
|
21 |
-
|
22 |
-
- **Model type:** BART fine-tuned for abstractive summarization
|
23 |
-
- **Language:** English
|
24 |
-
- **Training data:** SAMSum dataset
|
25 |
-
- **Metrics:** ROUGE-1: 54.52, ROUGE-2: 30.88, ROUGE-L: 45.09
|
26 |
-
|
27 |
-
## Usage
|
28 |
-
|
29 |
-
```python
|
30 |
-
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
31 |
-
|
32 |
-
model = AutoModelForSeq2SeqLM.from_pretrained("ASQWADSda/meeting-summarizer")
|
33 |
-
tokenizer = AutoTokenizer.from_pretrained("ASQWADSda/meeting-summarizer")
|
34 |
-
|
35 |
-
meeting_transcript = '''
|
36 |
-
John: Welcome everyone to our weekly product planning meeting.
|
37 |
-
Sarah: Thanks John. I've prepared the updated roadmap for Q3.
|
38 |
-
...
|
39 |
-
'''
|
40 |
-
|
41 |
-
inputs = tokenizer(meeting_transcript, max_length=512, truncation=True, return_tensors="pt")
|
42 |
-
outputs = model.generate(inputs.input_ids, max_length=128, num_beams=4)
|
43 |
-
summary = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
44 |
-
print(summary)
|
45 |
-
```
|
46 |
-
|
47 |
-
## Limitations
|
48 |
-
|
49 |
-
This model works best on conversational meeting transcripts with clear speaker attribution.
|
50 |
-
It may not perform as well on formal speeches or presentations.
|
|
|
14 |
|
15 |
# Meeting-Mind: Meeting Summarization Model
|
16 |
|
17 |
+
This model was fine-tuned on the SAMSum dataset to generate concise summaries of meeting transcripts.
|
18 |
It uses a BART base model with 2 epochs of training.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|